Underwater vehicle acoustic ranging-only intelligent positioning method based on four base stations
Through the combination of the non-coplanar configuration of the four base stations and the radial basis neural network, the problem of high-precision real-time navigation of underwater vehicles is solved, and the rapid positioning of underwater rescue tasks is achieved.
Patent Information
- Application Number
- CN202411962311.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-30
- Publication Date
- 2025-04-18
AI Technical Summary
Underwater vehicles cannot use GPS or visual navigation systems for high-precision navigation and positioning. Traditional acoustic positioning methods have problems such as ambiguity and high calculation volume, and cannot meet the real-time and accuracy requirements of underwater rescue tasks such as lakes and rivers.
The four base stations are arranged non-coplanar, combined with acoustic beacons and radial basis neural networks, and the distance measurement model between the underwater vehicle and the fixed base station is established through the field value culling and smoothing algorithm of the distance measurement value to achieve real-time and high-precision navigation and positioning.
It realizes fast and high-precision autonomous navigation of underwater vehicles in waters such as lakes and rivers, reduces the requirements of computing resources, and solves the ambiguity and real-time problems in traditional methods.
Smart Images
Figure CN119846637A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of underwater vehicle autonomous navigation, and in particular relates to an underwater vehicle ranging-only intelligent positioning method based on four base stations. Background Art
[0002] The demand for underwater vehicles in many tasks such as water quality monitoring and underwater life search and rescue is becoming more and more vigorous. However, due to signal transmission problems, underwater vehicles cannot use global satellite navigation systems such as GPS and Beidou to complete their own navigation and positioning, and the low-cost low-precision inertial navigation positioning errors will diverge quickly, and cannot meet the high-precision requirements for navigation and positioning in search and rescue missions. At the same time, due to the influence of underwater visibility, commonly used visual navigation methods such as SLAM cannot be used underwater. The method based on acoustic measurement has less propagation attenuation underwater and can be transmitted over long distances, so acoustic positioning has become one of the most important ways for underwater vehicle navigation. Generally, in the field of underwater rescue in lakes and rivers, the real-time requirements for navigation and positioning are very high. The positioning method of arranging a single or two acoustic beacon base stations requires multiple measurement accumulations and special maneuvering assistance such as S-type to complete navigation and positioning, which cannot meet the real-time requirements; the navigation and positioning method of three base stations also requires special maneuvering assistance, and there are problems such as signal shielding or time-frequency errors that cannot achieve high precision. At the same time, existing solution methods all involve iterative solutions to nonlinear equations, which require a large amount of calculation and high computing power for the vehicle. Summary of the invention
[0003] In view of the deficiencies of the above-mentioned prior art, the purpose of the present invention is to provide an acoustic ranging-only navigation method for underwater vehicles based on four base stations for underwater search and rescue missions in lakes, rivers, etc. The method has a reasonable design and is simple to operate. It does not require high-precision inertial navigation and other positioning systems. It only uses pre-deployed fixed-position acoustic beacon base stations, and utilizes the transducer installed on the underwater vehicle to transmit the acoustic signal and the transponder on the underwater base station to return the signal to measure the relative distance, thereby realizing autonomous navigation and positioning of the underwater vehicle without the need for special maneuvering assistance.
[0004] The present invention provides an underwater vehicle acoustic ranging-only navigation method based on four base stations, comprising the following steps:
[0005] Step 1, establish a non-coplanar configuration and deploy four acoustic beacon fixed base stations;
[0006] Step 2, establish an algorithm model for eliminating outliers in the distance measurement between the underwater vehicle and the fixed base station, eliminate outliers by comparing the data of the previous and subsequent distance measurements, improve the measurement information, and obtain accurate distance measurement values;
[0007] Step 3, establishing a navigation and positioning algorithm model based on acoustic ranging between the underwater vehicle and four base stations, and inputting the ranging value to calculate the three-dimensional position vector of the underwater vehicle;
[0008] Step 4, establish a four-base station acoustic ranging navigation positioning solution model based on radial basis function neural network, input a data sequence containing relative position, speed and ranging value, output the real-time position vector of the underwater vehicle, and realize the real-time navigation positioning of the underwater vehicle.
[0009] Preferably, in step 1), the four-base station layout configuration for the underwater mission is designed such that three base stations are located in the same plane, forming three vertices of a triangle, and the fourth base station is placed outside the plane and is not coplanar with the other three base stations. The purpose of such design is to solve the ambiguity problem caused by coplanar base station ranging positioning, that is, the three-base station or two-base station ranging positioning mode will inevitably have multiple solutions, and accurate positioning cannot be achieved. It is necessary to rely on the aircraft to perform a large-scale S-shaped maneuver to remove the ambiguity, which is not conducive to rapid underwater rescue missions.
[0010] Preferably, in step 2), if the absolute value of the difference between the current distance measurement value and the distance measurement value at the previous moment exceeds the average displacement distance ||Δr||, the current distance measurement value is discarded and revised to the distance measurement value at the previous moment, expressed as:
[0011]
[0012] In the formula, j represents the measurement time,
[0013] Preferably, in step 2), a sliding window is introduced to smooth the ranging data, which is expressed as:
[0014]
[0015] Where i = 1, 2, 3, 4 represents the base station number, ρ i,j represents the distance between the underwater vehicle and base station #i.
[0016] Preferably, in step 3), the transducer of the underwater vehicle broadcasts an acoustic signal at a fixed frequency, and the base station immediately returns a response signal after receiving the signal. After receiving the returned signal, the transducer of the underwater vehicle records the time difference between sending and receiving the acoustic signal, and calculates the distance measurement according to the time difference: Among them, v is the speed of sound, Δt is the time difference between signal transmission and reception. Four distance measurements of four base stations can be obtained at a single time, and the ranging relationship between the four underwater vehicles and the base stations can be obtained, and the unknown number - the three-dimensional position vector of the underwater vehicle - can be fitted and solved.
[0017] As a preference, the ranging relationship between the underwater vehicle and the base station is expressed as:
[0018]
[0019] Where i = 1, 2, 3, 4 represents the base station number, j represents the measurement time; P vj represents the position vector of the underwater vehicle at time j, x j ,y j and z j is the three-axis component corresponding to the spacecraft at time j; P bi represents the position vector of the i-th base station, X bi , Y bi and Z bi is the three-axis component corresponding to the i-th base station; n ρ is the measurement noise.
[0020] Preferably, the solution model in step 4) includes a data generator and a radial basis function neural network, wherein the data generator generates a relative position, speed and four-base station distance measurement data sequence in the base station measurement coordinate system according to the uniform straight line, turning, floating and diving motion primitives and dynamic models commonly used by underwater vehicles, and the generated training data is normalized and preprocessed before inputting into a three-layer feedforward radial basis function neural network with four inputs and three outputs. The gradient descent optimization algorithm of the neural network adopts the SGD algorithm, and the normalized data is used as the input of the radial basis function neural network for offline training, and a mapping network model from four distance measurement information to the three-dimensional position coordinates of the underwater vehicle is established, so as to realize offline training and online solution.
[0021] Preferably, the neural network comprises a three-layer feedforward radial basis neural network including an input layer, a hidden layer and an output layer, and the hidden layer is represented by a radial basis function as follows: Among them, x i is the i-th sample, m j is the center of the jth hidden layer node, and is σ j is the width parameter of the hidden layer.
[0022] As a preferred method, the SGD algorithm is used as the stochastic gradient descent optimization algorithm of the neural network, the number of neural network layers is 3, and the mean square error of the underwater vehicle position coordinates output by the radial basis neural network and its nominal value is used as the loss function of the network training, which is expressed as: Where n is the number of samples, P vjl is the nominal value of the spacecraft position coordinate, The location coordinate prediction value output by the network model.
[0023] As a preference, the normalization preprocessing is: Among them, x is the original data, The normalized data is used as the input of the radial basis function neural network for offline training, and a mapping network model from four distance measurement information to the three-dimensional position coordinates of the underwater vehicle is established; then the network model is deployed on the underwater vehicle, and the measured distance information is input into the network model to realize offline training and online position coordinate solution.
[0024] Beneficial effects of the present invention:
[0025] The present invention solves the ambiguity problem of ranging and positioning in the case of coplanar base stations or fewer base stations by designing four acoustic base stations in a non-coplanar tetrahedral configuration, thereby providing a signal source support for realizing high-precision and rapid positioning; utilizing the motion characteristics of the underwater vehicle, an algorithm model for eliminating wild values and smoothing interpolation of distance measurements between the underwater vehicle and the fixed base station is established; a radial basis function neural network is designed, and a four-base station acoustic ranging navigation and positioning solution model is constructed, thereby realizing real-time, rapid, and high-precision solution of the underwater vehicle. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a diagram of the pre-arranged non-coplanar configuration of four acoustic beacon base stations;
[0027] Figure 2 It is a schematic diagram of the acoustic measurement geometry of four base stations of an underwater vehicle;
[0028] Figure 3 It is a radial basis neural network structure design diagram;
[0029] Figure 4 is the motion trajectory diagram of the underwater vehicle;
[0030] Figure 5 It is the actual distance measurement curve of underwater vehicle and four base stations;
[0031] Figure 6 It is the underwater vehicle-four base station ranging error curve;
[0032] Figure 7 It is the three-axis positioning error curve of underwater vehicle. DETAILED DESCRIPTION
[0033] In order to facilitate the understanding of those skilled in the art, the present invention is further described below in conjunction with embodiments and drawings. The contents mentioned in the implementation modes are not intended to limit the present invention.
[0034] The present invention proposes an underwater vehicle acoustic ranging-only intelligent positioning method based on four base stations. In view of the problems that in the current underwater rescue scenarios in waters such as lakes and rivers, navigation methods such as satellite navigation systems and visual SLAM navigation cannot work, traditional acoustic positioning methods are prone to fuzzy solutions, require multiple measurements and accumulation for positioning and cannot meet real-time requirements, and have high requirements on airborne computing resources; the method of the present invention establishes four acoustic beacon base stations in a non-coplanar configuration, establishes a wild value elimination and smooth interpolation algorithm model for distance measurement values between the underwater vehicle and the fixed base station to improve the quality of measurement information, establishes a four-base station acoustic ranging positioning model, and introduces a radial basis function neural network to design a positioning solution algorithm, so as to achieve real-time, fast and high-precision solution of the underwater vehicle.
[0035] The specific instructions are as follows:
[0036] Step 1: Design the four-base station deployment configuration for underwater missions.
[0037] like Figure 1 As shown, the four base stations are not completely coplanar. In this embodiment, the four base stations adopt a non-coplanar configuration: specifically, three base stations are located in the same plane, forming the three vertices of a triangle; and the fourth base station is placed outside the plane, not coplanar with the remaining three base stations, and the four base stations form a tetrahedron configuration. The purpose of this design is to solve the ambiguity problem caused by coplanar base station ranging and positioning, that is, the three-base station or two-base station ranging and positioning mode will inevitably cause the problem of multiple solutions due to the lack of azimuth measurement, and cannot achieve accurate positioning. It is necessary to rely on the aircraft to perform large-scale S-shaped maneuvers to remove ambiguity, which is not conducive to rapid underwater rescue missions. At the same time, if the configuration of the four base stations is not appropriate, there will also be multiple solutions, and the true position of the aircraft cannot be determined.
[0038] Step 2: Establish an algorithm model for eliminating outliers in the distance measurement between the underwater vehicle and the fixed base station. The model is as follows:
[0039] a) Since the underwater vehicle moves slowly, outliers are eliminated by comparing the previous and subsequent ranging data. At the same time, in order to solve the adaptability problem caused by the fixed threshold of the traditional outlier elimination method, the motion characteristics of the underwater vehicle are introduced here to set an adaptive threshold. Specifically, if the absolute value of the difference between the current ranging value and the ranging value at the previous moment exceeds the average displacement distance ||Δr||, the current ranging value is eliminated and revised to the ranging value at the previous moment:
[0040]
[0041] Where j represents the measurement time, Δt = t j+1 -t j , v j and v j+1 They are tj and t j+1 The speed of the underwater vehicle at time
[0042] b) In order to smooth the influence of random measurement errors, a sliding window is introduced to smooth the ranging data:
[0043]
[0044] Where i = 1, 2, 3, 4 represents the base station number, ρ i,j represents the distance between the underwater vehicle and base station #i.
[0045] Step 3: Establish a navigation and positioning algorithm model based on acoustic ranging between underwater vehicles and four base stations. The model is as follows:
[0046] The distance measurement between the underwater vehicle and the base station satisfies the following equation:
[0047]
[0048] Where i = 1, 2, 3, 4 represents the base station number, j represents the measurement time; P vj represents the position vector of the underwater vehicle at time j, x j ,y j and z j They are the corresponding three-axis components; P bi represents the position vector of the i-th base station, X bi , Y bi and Z bi They are the corresponding three-axis components; n ρ is the measurement noise.
[0049] like Figure 2 As shown, the transducer of the underwater vehicle broadcasts an acoustic signal at a fixed frequency. After receiving the signal, the base station immediately returns a response signal. After receiving the returned signal, the transducer of the underwater vehicle obtains the distance measurement based on the phase difference.
[0050] Record the time difference between sending and receiving the acoustic signal, and calculate the distance measurement based on the time difference:
[0051]
[0052] Where v is the speed of sound and Δt is the time difference between sending and receiving the signal.
[0053] A single measurement can obtain four distance measurements, so it is possible to establish a system of equations with the three-dimensional position vector of the underwater vehicle at a certain time j as the unknown:
[0054]
[0055] Step 4: Establish a four-base station acoustic ranging navigation and positioning solution model based on radial basis function neural network to achieve high-precision and lightweight computational navigation and positioning of underwater vehicles.
[0056] a) Define the underwater vehicle ranging and positioning training data generator. Generate the relative position, speed and four base station distance measurement data sequence in the base station measurement coordinate system according to the motion primitives and dynamic models commonly used by underwater vehicles, such as uniform straight line, turning, floating, diving, etc.
[0057] b) Perform normalization preprocessing on the generated training data:
[0058]
[0059] Among them, x is the original data, The data are normalized.
[0060] c) Define the radial basis neural network. Design Figure 3 The three-layer feedforward radial basis function neural network with four inputs and three outputs shown in the figure is composed of an input layer, a hidden layer and an output layer. Preferably, the hidden layer adopts the radial basis function as follows:
[0061]
[0062] Among them, x i is the i-th sample, m j is the center of the jth hidden layer node, and is σ j is the width parameter of the hidden layer.
[0063] This embodiment uses the SGD algorithm as the stochastic gradient descent optimization algorithm of the neural network, the number of neural network layers is 3, and the mean square error of the underwater vehicle position coordinates output by the radial basis neural network and its nominal value is used as the loss function of the network training, which is as follows:
[0064]
[0065] Where n is the number of samples, P vjl is the nominal value of the spacecraft position coordinate, The location coordinate prediction value output by the network model.
[0066] The normalized data in step 4.b is used as the input of the radial basis neural network for offline training, and a mapping network model from the four distance measurement information to the three-dimensional position coordinates of the underwater vehicle is established. The network model is then deployed on the underwater vehicle, and the measured distance information is input into the network model to achieve offline training and fast high-precision position coordinate solution for online use.
[0067] To further verify the beneficial effects of this application, Figures 3 to 7 The example verification of this application is described, and the following calculation conditions and technical parameters are set:
[0068] 1) Deploy four non-coplanar acoustic base stations, and establish the northeast sky navigation coordinate system with base station #1 as the origin. The coordinates of the four base stations are:
[0069] P1=[0,0,0]; P2=[0,100,0]; P3=[120,0,1]; P4=[120,100,-5]m;
[0070] 2) The initial position and speed of the underwater vehicle in the navigation coordinate system are:
[0071] [2m,0m,-0.5m,0m / s,0m / s,0m / s];
[0072] 3) The motion forms of underwater vehicles include uniform acceleration, uniform speed, uniform deceleration, turning, floating, diving, etc.;
[0073] 4) The normal ranging error is a random error with a mean of zero and a mean square error of 0.1 m, and the abnormal ranging error is a random error with a mean square error of 1;
[0074] Based on the underwater vehicle ranging only intelligent positioning method of the present invention and the above-mentioned calculation conditions and technical parameters, a numerical simulation verification is carried out, and the simulation time is 600s. Figures 3 to 7 The figures show the motion trajectory of the underwater vehicle, the nominal four-base station ranging curve, the ranging error curve and the three-axis positioning error curve. It can be seen from the curves in the figure that the method of the present invention can well reduce the influence of measurement abnormal values, and the designed non-coplanar measurement base station configuration can solve the measurement ambiguity problem and achieve high-precision positioning. The positioning error is generally within 0.1m.
[0075] Therefore, by adopting the method of the present invention, the rapid and high-precision positioning task of the underwater vehicle can be achieved by relying only on the response ranging of four non-coplanar acoustic base stations.
[0076] The present invention has many specific application paths. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements can be made without departing from the principle of the present invention. These improvements should also be regarded as the protection scope of the present invention.
Claims
1. An acoustic ranging-only navigation method for underwater vehicles based on four base stations, characterized in that: The steps include: Step 1, establish a non-coplanar configuration and deploy four acoustic beacon fixed base stations; Step 2, establish an algorithm model for eliminating outliers in the distance measurement between the underwater vehicle and the fixed base station, eliminate outliers by comparing the data of the previous and subsequent distance measurements, improve the measurement information, and obtain accurate distance measurement values; Step 3, establishing a navigation and positioning algorithm model based on acoustic ranging between the underwater vehicle and four base stations, and inputting the ranging value to calculate the three-dimensional position vector of the underwater vehicle; Step 4, establish a four-base station acoustic ranging navigation positioning solution model based on radial basis function neural network, input a data sequence containing relative position, speed and ranging value, output the real-time position vector of the underwater vehicle, and realize the real-time navigation positioning of the underwater vehicle.
2. The underwater vehicle acoustic ranging-only navigation method according to claim 1, characterized in that: In step 1, the non-coplanar configuration of four base stations is that three base stations are located in the same plane, forming three vertices of a triangle, and the fourth base station is placed outside the plane and is not coplanar with the remaining three base stations.
3. The underwater vehicle acoustic ranging-only navigation method according to claim 1, characterized in that: In step 2, if the absolute value of the difference between the current distance measurement value and the distance measurement value at the previous moment exceeds the average displacement distance ||Δr||, the current distance measurement value is discarded and revised to the distance measurement value at the previous moment, expressed as: In the formula, j represents the measurement time, 4. The underwater vehicle acoustic ranging-only navigation method according to claim 3, characterized in that: In step 2, a sliding window is introduced to smooth the ranging data, which is expressed as: Where i = 1, 2, 3, 4 represents the base station number, ρ i,j represents the distance between the underwater vehicle and base station #i.
5. The underwater vehicle acoustic ranging-only navigation method according to claim 1, characterized in that: In step 3, the transducer of the underwater vehicle broadcasts an acoustic signal at a fixed frequency. After receiving the signal, the base station immediately returns a response signal. After receiving the returned signal, the transducer of the underwater vehicle records the time difference between sending and receiving the acoustic signal, and calculates the distance measurement based on the time difference: Among them, v is the speed of sound, Δt is the time difference between signal transmission and reception. Four distance measurements of four base stations can be obtained at a single time, and the ranging relationship between the four underwater vehicles and the base stations can be obtained, and the unknown number - the three-dimensional position vector of the underwater vehicle - can be fitted and solved.
6. The underwater vehicle acoustic ranging-only navigation method according to claim 5, characterized in that: The distance measurement relationship between the underwater vehicle and the base station is expressed as: Where i = 1, 2, 3, 4 represents the base station number, j represents the measurement time; P vj represents the position vector of the underwater vehicle at time j, x j ,y j and z j is the three-axis component corresponding to the spacecraft at time j; P bi represents the position vector of the i-th base station, X bi , Y bi and Z bi is the three-axis component corresponding to the i-th base station; n ρ is the measurement noise.
7. The underwater vehicle acoustic ranging-only navigation method according to any one of claims 1 to 6, characterized in that: In step 4, the solution model includes a data generator and a radial basis neural network. The data generator generates a relative position, speed and four-base station distance measurement data sequence in the base station measurement coordinate system according to the uniform straight line, turning, floating and diving motion primitives and dynamic models commonly used by underwater vehicles. The generated training data is normalized and preprocessed and then input into a three-layer feedforward radial basis neural network with four inputs and three outputs. The gradient descent optimization algorithm of the neural network adopts the SGD algorithm, and the normalized data is used as the input of the radial basis neural network for offline training, and a mapping network model from four distance measurement information to the three-dimensional position coordinates of the underwater vehicle is established to achieve offline training and online solution.
8. The underwater vehicle acoustic ranging-only navigation method according to claim 7, characterized in that: In step 4, the neural network includes a three-layer feedforward radial basis neural network including an input layer, a hidden layer and an output layer, and the hidden layer is represented by a radial basis function as follows: Among them, x i is the i-th sample, m j is the center of the jth hidden layer node, and is σ j is the width parameter of the hidden layer.
9. The underwater vehicle acoustic ranging-only navigation method according to claim 7, characterized in that: The SGD algorithm is used as the stochastic gradient descent optimization algorithm of the neural network. The number of neural network layers is 3. The mean square error of the underwater vehicle position coordinates output by the radial basis neural network and its nominal value is used as the loss function of the network training, which is expressed as: Where n is the number of samples, P vjl is the nominal value of the spacecraft position coordinate, The location coordinate prediction value output by the network model.
10. The underwater vehicle acoustic ranging-only navigation method according to claim 7, characterized in that: The normalization preprocessing is: Among them, x is the original data, The normalized data is used as the input of the radial basis function neural network for offline training, and a mapping network model from four distance measurement information to the three-dimensional position coordinates of the underwater vehicle is established; then the network model is deployed on the underwater vehicle, and the measured distance information is input into the network model to realize offline training and online position coordinate solution.