Unmanned surface vessel path planning method and system

Abstract

The invention discloses an unmanned surface vessel path planning method and system, particularly relates to the technical field of path planning, and is used for solving the problem of course deviation caused by difficulty in avoiding a local water area wave interference area in real time in the prior art. The method comprises the following steps: acquiring fluctuation data of a target water area, and performing multi-source sensor fusion processing to generate basic data for identifying a wave interference area; a deep neural network algorithm is adopted to identify an energy gathering area, and wave interference intensity distribution is output; path deviation analysis is carried out based on the wave interference intensity distribution, local risk factors are generated, and an avoidance threshold value is set; performing iterative planning on a preset route by using a particle swarm optimization algorithm, and dynamically adjusting route nodes; correcting the real-time position and attitude information through a Kalman filtering algorithm, and outputting a final navigation path meeting an avoidance threshold value; the path planning capability of the unmanned surface vessel in a complex water area is improved, and safe and efficient navigation is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and more specifically, to a method and system for unmanned surface vessel path planning. Background Art

[0002] As an advanced unmanned water platform, unmanned surface vessels are widely used in fields such as ocean exploration, environmental monitoring, and water patrol. Their operation path planning is an important technical link to ensure the efficiency and safety of mission execution. Existing path planning technologies usually optimize paths based on fixed rules or simple environmental characteristics, and it is difficult to fully consider complex factors in dynamic marine environments, such as the fluctuation characteristics of local waters. Especially in terms of multi-source data fusion and real-time path adjustment, the response speed and accuracy of traditional technologies still have a lot of room for improvement. With the expansion of the application scenarios of unmanned surface vessels, their path planning capabilities need to be further improved to adapt to more complex water conditions.

[0003] In the existing technology, local waters may form dynamic energy concentration areas or turbulent areas due to wave interference, and it is difficult to identify and avoid these high-risk areas in real time, which will cause the unmanned surface vessel's navigation path to deviate, increase energy consumption, and even the risk of loss of control. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an unmanned surface vessel path planning method and system to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A path planning method for an unmanned surface vehicle comprises the following steps:

[0007] S1: Obtain the wave data of the target water area, and perform multi-source sensor fusion processing on the wave data to generate basic data for identifying the local wave interference area;

[0008] S2: Use deep neural network algorithm to identify energy concentration areas on basic data and output wave interference intensity distribution;

[0009] S3: Analyze the path deviation of the historical navigation trajectory of the unmanned surface vessel based on the wave interference intensity distribution, generate local risk factors and set avoidance thresholds;

[0010] S4: Use the particle swarm optimization algorithm to iteratively plan the preset route of the unmanned surface vehicle, and adjust the updated node coordinates according to the local risk factor during each iteration;

[0011] S5: The real-time position and attitude information of the unmanned surface vehicle is dynamically corrected through the Kalman filter algorithm, and the final navigation path that meets the avoidance threshold is output.

[0012] In a preferred embodiment, the wave data of the target water area is obtained, and the wave data is subjected to multi-source sensor fusion processing to generate basic data for identifying the local wave interference area, specifically including:

[0013] Use multiple sensors installed on the unmanned surface vessel to obtain fluctuation data of the target water area, including radar sensors, sonar sensors and laser rangefinders;

[0014] Temporally and spatially align the water surface wave height data obtained by the radar sensor, the underwater wave amplitude data obtained by the sonar sensor, and the water surface texture data obtained by the laser rangefinder;

[0015] Based on the weighted average fusion algorithm, multi-source sensor fusion processing is performed on the water surface wave height data, underwater wave amplitude data and water surface texture data after time and space alignment to generate fusion data for characterizing local wave characteristics;

[0016] Information including the wave frequency, wave height change rate and wave direction is extracted from the fused data to generate basic data for identifying local wave interference areas.

[0017] In a preferred embodiment, a deep neural network algorithm is used to identify energy concentration areas on basic data, and the wave interference intensity distribution is output, specifically including:

[0018] The basic data are used as input data of the deep neural network model. The basic data include the distribution of fluctuation frequency, the distribution of wave height change trend and the distribution of fluctuation propagation direction.

[0019] Construct a multi-layer deep neural network model, including an input layer, multiple hidden layers and an output layer, wherein the input layer is used to receive the multi-dimensional features of the basic data, the hidden layer includes multiple fully connected layers and activation functions, the activation function uses a rectified linear unit, and the output layer is used to generate the spatial distribution probability of the energy concentration area;

[0020] The deep neural network model is trained with the training data including the wave characteristic data based on the actual sea area and the corresponding energy concentration area label data;

[0021] The basic data to be analyzed is input into the trained deep neural network model, and the energy concentration area distribution probability map of the target water area is obtained through model calculation;

[0022] The wave interference intensity is calculated according to the energy concentration area distribution probability map to generate wave interference intensity distribution data, which includes the wave interference intensity value of each spatial point.

[0023] In a preferred embodiment, the wave interference intensity is calculated according to the energy concentration area distribution probability map to generate the wave interference intensity distribution data, specifically:

[0024] According to the probability map of energy concentration area distribution, the wave interference intensity value of each spatial point is calculated: I = P·F; where I represents the wave interference intensity, P represents the energy concentration probability value, and F represents the wave frequency of the corresponding spatial point;

[0025] Generation of wave interference intensity distribution data: Filter the wave interference intensity values ​​to remove possible noise interference in the probability map; integrate the filtered wave interference intensity values ​​into distribution data to form the wave interference intensity distribution of each spatial point in the target waters.

[0026] In a preferred embodiment, path deviation analysis is performed on the historical navigation trajectory of the unmanned surface vessel based on the wave interference intensity distribution, a local risk factor is generated and an avoidance threshold is set, specifically including:

[0027] The wave interference intensity distribution data is superimposed and analyzed with the historical navigation track of the unmanned surface vessel, wherein the historical navigation track data includes the spatial coordinates and time series information of the navigation path;

[0028] Calculate a deviation vector based on the actual path and the planned path of the historical navigation trajectory of the unmanned surface vessel;

[0029] Based on the wave interference intensity value of each spatial point in the wave interference intensity distribution data, the weight of the deviation vector is corrected to obtain a path deviation correction value;

[0030] Map the path deviation correction value to the spatial point of the local water area to generate the local risk factor corresponding to each spatial point. The local risk factor is positively correlated with the path deviation correction value.

[0031] The avoidance threshold is set according to the distribution characteristics of the local risk factor. The avoidance threshold is the condition that the route needs to be adjusted when the local risk factor is higher than the preset value;

[0032] The local risk factors are normalized to form normalized risk factor distribution data.

[0033] In a preferred embodiment, the preset route of the unmanned surface vehicle is iteratively planned using a particle swarm optimization algorithm, and the updated node coordinates are adjusted according to the local risk factor during each iteration, specifically including:

[0034] The preset route of the unmanned surface vehicle is discretized into multiple nodes, each node contains the corresponding spatial coordinate information, and the node is used as the initial solution of the particle swarm optimization algorithm;

[0035] Initialize the relevant parameters of the particle swarm optimization algorithm, including the number of particles, particle speed, particle search space range, inertia weight, individual optimal weight and group optimal weight;

[0036] According to the local risk factor distribution data, a corresponding risk factor weight is assigned to each node of the route, and the spatial coordinates of the node and the risk factor weight are used as the input of the fitness evaluation function of the particle swarm optimization algorithm;

[0037] In each iteration, the particle swarm optimization algorithm is used to calculate the fitness value of the particle, which is evaluated based on the spatial coordinate offset of the node and the risk factor weight;

[0038] Dynamically update the speed and position of particles. The speed of particles is used to control the adjustment amplitude of node coordinates during the update process, and the current position of particles is adjusted according to the updated speed to gradually approach the global optimal solution.

[0039] After each iteration, the fitness evaluation criteria are dynamically adjusted according to the current local risk factor distribution data, so that the route nodes are gradually optimized to the low risk factor area;

[0040] When the convergence condition of the particle swarm optimization algorithm is met, the optimized route node coordinates are output as the updated route of the unmanned surface vehicle.

[0041] In a preferred embodiment, the real-time position and attitude information of the unmanned surface vehicle is dynamically corrected by a Kalman filter algorithm, and a final navigation path that meets the avoidance threshold is output, specifically including:

[0042] Acquire real-time sensor data of the unmanned surface vehicle, including position coordinates, attitude angle and movement speed;

[0043] Construct a Kalman filter model, set the state vector as the real-time position and attitude information of the unmanned surface vehicle, and the observation vector as the data collected by the sensor;

[0044] According to the real-time sensor data and the predicted value of the Kalman filter model, the error covariance matrix and Kalman gain are calculated to correct the state vector;

[0045] Dynamically update the state vector and map the corrected position and attitude information to the actual running trajectory of the unmanned surface vessel;

[0046] A navigation path that meets the avoidance threshold is generated based on the corrected real-time position and attitude information.

[0047] On the other hand, the present invention provides an unmanned surface vessel path planning system, including a data acquisition and fusion module, an energy identification and analysis module, a path deviation evaluation module, a route optimization and planning module, and a dynamic path correction module;

[0048] Data acquisition and fusion module: obtains the wave data of the target water area, and performs multi-source sensor fusion processing on the wave data to generate basic data for identifying the local wave interference area;

[0049] Energy identification and analysis module: uses deep neural network algorithm to identify energy concentration areas based on basic data and outputs wave interference intensity distribution;

[0050] Path deviation assessment module: Analyze the path deviation of the historical navigation trajectory of the unmanned surface vessel based on the wave interference intensity distribution, generate local risk factors and set avoidance thresholds;

[0051] Route optimization planning module: It uses the particle swarm optimization algorithm to iteratively plan the preset route of the unmanned surface vehicle, and adjusts the updated node coordinates according to the local risk factor during each iteration;

[0052] Dynamic correction path module: The real-time position and attitude information of the unmanned surface vessel are dynamically corrected through the Kalman filter algorithm, and the final navigation path that meets the avoidance threshold is output.

[0053] The technical effects and advantages of the unmanned surface vessel path planning method and system of the present invention are as follows:

[0054] 1. By combining multi-source sensor fusion processing of target water area fluctuation data with deep neural network algorithms, it is possible to accurately identify local wave interference areas and their energy concentration distribution, thereby providing high-precision basic data support for subsequent path planning. Compared with traditional technologies that rely only on a single sensor or rule-based algorithm, the method of the present invention greatly improves the perception of local fluctuation characteristics and the accuracy of data processing, ensuring more efficient real-time monitoring and identification of dynamic risk areas in complex water environments.

[0055] 2. Based on route optimization of local risk factors and dynamic correction of Kalman filtering, the present invention realizes a full-process dynamic adjustment mechanism from route planning to actual operation. In the process of path optimization, the particle swarm optimization algorithm is used to iteratively update the route nodes, effectively avoiding local high-risk areas; at the same time, the real-time correction of the Kalman filter algorithm can further improve the navigation accuracy of the unmanned surface vessel and reduce the risk of excessive energy consumption or mission failure caused by path deviation. The path planning method of the present invention can not only adapt to the dynamic water environment, but also achieve safer and more efficient navigation, providing reliable guarantee for the execution of tasks by unmanned surface vessels in complex waters. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A schematic diagram of a path planning method for an unmanned surface vehicle according to the present invention;

[0057] Figure 2 The present invention is a schematic structural diagram of an unmanned surface vessel path planning system. DETAILED DESCRIPTION

[0058] 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.

[0059] Embodiment 1: Figure 1 The present invention provides a path planning method for an unmanned surface vehicle, which comprises the following steps:

[0060] S1: Obtain the wave data of the target water area, and perform multi-source sensor fusion processing on the wave data to generate basic data for identifying the local wave interference area.

[0061] S2: Use deep neural network algorithm to identify energy concentration areas on basic data and output wave interference intensity distribution.

[0062] S3: Analyze the path deviation of the historical navigation trajectory of the unmanned surface vessel based on the wave interference intensity distribution, generate local risk factors and set avoidance thresholds.

[0063] S4: The preset route of the unmanned surface vessel is iteratively planned using the particle swarm optimization algorithm, and the updated node coordinates are adjusted according to the local risk factor during each iteration.

[0064] S5: The real-time position and attitude information of the unmanned surface vehicle is dynamically corrected through the Kalman filter algorithm, and the final navigation path that meets the avoidance threshold is output.

[0065] Obtain the wave data of the target waters, and perform multi-source sensor fusion processing on the wave data to generate basic data for identifying the local wave interference area, including:

[0066] Use multiple sensors installed on unmanned surface vessels to obtain fluctuation data of the target waters, including radar sensors, sonar sensors and laser rangefinders:

[0067] Radar sensors are used to obtain surface wave height data and monitor the amplitude changes of water surface fluctuations in real time. The data they obtain contains the timing information of wave height. Sonar sensors are used to collect underwater fluctuation amplitude data and measure the frequency and intensity of underwater fluctuations through the reflection and refraction characteristics of sound waves underwater. Laser rangefinders are mainly used to capture water surface texture data and obtain the texture characteristics of local water surface fluctuations, including information such as changes in water surface curvature.

[0068] Sensors are installed at different locations on the unmanned surface vessel to ensure comprehensive coverage of the target water area fluctuation data and ensure the diversity of collected data. Each sensor works independently, and the acquired data is transmitted to the data processing unit through the sensor module.

[0069] Temporally and spatially align the water surface wave height data obtained by the radar sensor, the underwater wave amplitude data obtained by the sonar sensor, and the water surface texture data obtained by the laser rangefinder:

[0070] First, the collected fluctuation data are time-aligned according to the sampling time of the sensor so that the data of different sensors at the same time point can match. The time alignment process is implemented through an interpolation algorithm to fill the time difference caused by the sampling frequency difference between different sensors; then, according to the position distribution of the sensors on the unmanned surface vessel, the data are spatially aligned, and the surface wave height data of the radar sensor, the underwater fluctuation amplitude data of the sonar sensor, and the surface texture data of the laser rangefinder are uniformly mapped to the same spatial coordinate system. Spatial alignment is achieved through a geometric correction method based on coordinate transformation to ensure that the spatial positions corresponding to the data of each sensor are consistent.

[0071] The weighted average fusion algorithm is used to perform multi-source sensor fusion processing on the water surface wave height data, underwater wave amplitude data and water surface texture data after time and space alignment to generate fusion data for characterizing local wave characteristics:

[0072] In the fusion process, each sensor is assigned a weight according to its ability to perceive wave characteristics. For example, the radar sensor has a higher weight because it has a higher measurement accuracy for the wave height on the water surface; the sonar sensor is second because it has a more accurate measurement of the underwater wave amplitude; the laser rangefinder has the lowest weight, but still provides supplementary information on the water surface texture; after the weights are assigned, the wave height data, wave amplitude data, and water surface texture data obtained by each sensor are weighted and summed to calculate the corresponding fusion result at each time point; through the processing of the fusion algorithm, the fusion data representing the local wave characteristics is obtained. This data combines the advantages of radar, sonar, and laser rangefinders, and provides comprehensive basic data support for the identification of subsequent wave interference areas.

[0073] Extract information including wave frequency, wave height change rate and wave direction from the fused data to generate basic data for identifying local wave interference areas:

[0074] The frequency domain characteristics of the fused data are analyzed using Fourier transform to calculate the fluctuation frequency of the target water area and identify the main frequency components of the local fluctuations.

[0075] Based on the time series of wave height data, the wave height change rate is calculated, and the rate of change of wave height over time is extracted by the difference method to identify the severity of the wave intensity.

[0076] Finally, according to the changing trend of the spatial coordinates in the fused data, the wave direction information is extracted to determine the main and secondary directions of wave propagation, which facilitates the identification of possible interference areas.

[0077] The extracted fluctuation frequency, wave height change rate and fluctuation direction information are integrated to generate basic data that characterize the local fluctuation characteristics.

[0078] The basic data include the fluctuation frequency distribution, wave height change trend distribution and wave propagation direction distribution of each spatial point in the target waters; it can be used to identify wave interference areas and provide basic input for energy concentration area identification and path deviation analysis in subsequent steps.

[0079] The deep neural network algorithm is used to identify the energy concentration area of ​​the basic data and output the wave interference intensity distribution, including:

[0080] The basic data is used as the input data of the deep neural network model. The basic data includes the distribution of fluctuation frequency, the distribution of wave height change trend and the distribution of fluctuation propagation direction:

[0081] Fluctuation frequency distribution: represents the fluctuation frequency information at different locations of the target water area, reflecting the dynamic changes in the fluctuation characteristics of the water area.

[0082] Wave height change trend distribution: It indicates the rate at which wave height in the water area changes over time, and describes the severity of the fluctuation and the changing trend of wave energy intensity.

[0083] Wave propagation direction distribution: indicates the propagation direction of waves in the target waters, and is used to identify the possible main and secondary directions of wave energy accumulation.

[0084] The basic data is preprocessed to ensure that the data format matches the model input requirements, and is input into the deep neural network model in the form of a tensor, where the dimension of the tensor corresponds to the spatiotemporal distribution characteristics of the data.

[0085] Construct a multi-layer deep neural network model, including an input layer, multiple hidden layers and an output layer. The input layer is used to receive the multi-dimensional features of the basic data. The hidden layer includes multiple fully connected layers and activation functions. The activation function uses a rectified linear unit. The output layer is used to generate the spatial distribution probability of the energy concentration area:

[0086] Input layer: used to receive multi-dimensional features of basic data. The input data is represented in the form of tensors, including spatial points, time steps, and feature dimensions.

[0087] Hidden layer: includes multiple layers of fully connected layers and activation functions. The fully connected layer is used to extract the feature correlation of the data, and the number of nodes in each fully connected layer is dynamically adjusted according to the dimension of the input data. The activation function uses the rectified linear unit (ReLU), and its expression is: f(x) = max(0, x); where x is the value of the input data, and f(x) represents the output value after the activation function. The introduction of the rectified linear unit can effectively avoid the gradient vanishing problem and improve the efficiency of model training.

[0088] Output layer: used to generate the spatial distribution probability of the energy accumulation area. The output data is a three-dimensional tensor, which represents the energy accumulation probability value of each spatial point in the target water area.

[0089] The overall structure of the deep neural network model is adjusted according to the basic data characteristics of the actual target waters to ensure that the model has sufficient expressive power.

[0090] The deep neural network model is trained with the following training data: wave characteristic data based on the actual sea area and corresponding energy concentration area label data:

[0091] Collect the wave characteristic data of the target waters, including the historically measured wave frequency, wave height change trend and wave propagation direction. According to the actual observation results, mark the distribution of energy concentration areas to form corresponding regional label data.

[0092] The supervised learning method is adopted, with the fluctuation characteristic data as input and the energy concentration area label as output. The loss function is defined as the mean square error, which is expressed as: Where N represents the number of training samples, y i represents the true value of the i-th sample, represents the predicted value of the deep neural network model, and L represents the overall loss value.

[0093] The network parameters are optimized using the stochastic gradient descent algorithm, and the model weights are iteratively updated until the loss function converges.

[0094] Through training, the deep neural network model can learn the mapping relationship between basic data and the distribution of energy concentration areas.

[0095] The basic data to be analyzed is input into the trained deep neural network model, and the energy concentration area distribution probability map of the target water area is obtained through model calculation:

[0096] Input basic data, including fluctuation frequency distribution, wave height change trend distribution and fluctuation propagation direction distribution. The dimension of the data is consistent with the input layer of the model. The model transmits the input data layer by layer, calculates the feature weights through the full connection of the hidden layer, introduces nonlinear expression capabilities through the activation function, and finally generates a spatial distribution probability map in the output layer.

[0097] Each value in the energy concentration area distribution probability map represents the probability that the corresponding spatial point in the target water area belongs to the energy concentration area, and the probability value is between 0 and 1.

[0098] The wave interference intensity is calculated based on the energy concentration area distribution probability map to generate wave interference intensity distribution data, which includes the wave interference intensity value of each spatial point:

[0099] According to the probability map of energy concentration area distribution, the wave interference intensity value of each spatial point is calculated using the following formula: I = P·F; where I represents the wave interference intensity, P represents the energy concentration probability value, and F represents the fluctuation frequency of the corresponding spatial point. This formula combines the energy concentration probability and the fluctuation frequency to generate the wave interference intensity value.

[0100] Generation of wave interference intensity distribution data: Filter the wave interference intensity values ​​to remove possible noise interference in the probability map; integrate the filtered wave interference intensity values ​​into distribution data to form the wave interference intensity distribution of each spatial point in the target waters.

[0101] Based on the wave interference intensity distribution, the path deviation analysis of the historical navigation trajectory of the unmanned surface vessel is carried out to generate local risk factors and set avoidance thresholds, including:

[0102] The wave interference intensity distribution data is superimposed and analyzed with the historical navigation trajectory of the unmanned surface vessel, where the historical navigation trajectory data includes the spatial coordinates and time series information of the navigation path:

[0103] The wave interference intensity distribution data is a three-dimensional spatial distribution, describing the wave interference intensity value at each spatial point in the target waters. The historical navigation trajectory data includes the actual path of the unmanned surface vessel within a given navigation time, specifically expressed as the spatial coordinate sequence of the path and the corresponding time series information.

[0104] The spatial coordinates of the historical navigation trajectory are aligned with the coordinate system of the wave interference intensity distribution data to ensure that the two are superimposed in the same spatial reference system.

[0105] The result of the overlay analysis is a time series data set, in which each time point corresponds to the position on the actual path of the USV and the wave interference intensity value at that position.

[0106] The superimposed data are interpolated to ensure that the wave interference intensity distribution data can obtain accurate values ​​at each time point of the historical navigation trajectory.

[0107] According to the actual path and planned path of the historical navigation trajectory of the unmanned surface vessel, the deviation vector is calculated:

[0108] The planned path of the historical navigation trajectory is generated by the navigation model of the unmanned surface vessel under ideal conditions, and the actual path records the actual trajectory of the unmanned surface vessel after being affected by external factors.

[0109] The calculation formula of the deviation vector is: D=RP; where D represents the deviation vector, R represents the spatial coordinate vector of the actual path, and P represents the spatial coordinate vector of the planned path.

[0110] The magnitude of the deviation vector is represented by its modulus, which is calculated as: Among them, R x , R y , R z are the three-dimensional coordinate components of the actual path, P x , P y , P z are the three-dimensional coordinate components of the planned path respectively.

[0111] Based on the wave interference intensity value of each spatial point in the wave interference intensity distribution data, the weight of the deviation vector is corrected to obtain the path deviation correction value:

[0112] The correction process is completed by assigning a weight to each deviation vector, and the weight is determined by the wave interference intensity value. The weight correction formula is: D′=w·D; where D′ is the corrected deviation vector and w is the weight factor.

[0113] The weight factor is calculated by normalizing the wave interference intensity value: Where I is the wave interference intensity at a spatial point, I max is the maximum value in the wave interference intensity distribution data, and the value range of w is 0 to 1.

[0114] The path deviation correction value is mapped to the spatial point of the local water area to generate the local risk factor corresponding to each spatial point. The local risk factor is positively correlated with the path deviation correction value:

[0115] The calculation formula of the local risk factor is: R = |D′|; where R represents the local risk factor, and |D′| is the modulus length of the corrected deviation vector.

[0116] The avoidance threshold is set according to the distribution characteristics of the local risk factor. The avoidance threshold is the condition that the route needs to be adjusted when the local risk factor is higher than the preset value:

[0117] The distribution characteristics of local risk factors are obtained through statistical analysis, including the maximum value, minimum value and distribution range.

[0118] The avoidance threshold is set to trigger the route adjustment when the risk factor is higher than the preset value. For example, the calculation formula of the avoidance threshold is: T = R avg +k·σ; where T is the avoidance threshold, R avgis the average value of the local risk factor, σ is the standard deviation of the local risk factor, and k is the adjustment coefficient used to control the sensitivity of the threshold.

[0119] Normalize the local risk factors to form normalized risk factor distribution data:

[0120] The normalized data ranges from 0 to 1 and is used as a basis for adjustment in subsequent route planning.

[0121] The particle swarm optimization algorithm is used to iteratively plan the preset route of the unmanned surface vehicle, and the updated node coordinates are adjusted according to the local risk factors during each iteration, including:

[0122] The preset route of the unmanned surface vehicle is discretized into multiple nodes, each node contains the corresponding spatial coordinate information, and the node is used as the initial solution of the particle swarm optimization algorithm:

[0123] The preset route of the unmanned surface vehicle is represented by a continuous path. In order to facilitate the processing of the particle swarm optimization algorithm, the preset route needs to be discretized into multiple nodes. Each node contains the corresponding spatial coordinate information, including the three-dimensional coordinate position of the node.

[0124] The preset route is segmented according to a certain distance interval, and a node is generated at the starting point of each segment to ensure that the interval between nodes is small enough to improve the optimization accuracy; the spatial coordinate information of each node is used as the initial solution of the particle swarm optimization algorithm to provide basic input for subsequent optimization.

[0125] Through the above discretization process, the preset route is accurately decomposed into a set of finite number of nodes, which will be used as the initial positions of particles in the particle swarm optimization algorithm.

[0126] Initialize the relevant parameters of the particle swarm optimization algorithm, including the number of particles, particle speed, particle search space range, inertia weight, individual optimal weight and group optimal weight:

[0127] Determine the number of particles, that is, the total number of particles in the particle swarm. The number of particles must match the number of route nodes to ensure that each node can participate in the optimization calculation; initialize the particle speed, which is used to control the adjustment amplitude of the node coordinates in each iteration. The initial speed can be set to a random value to ensure the diversity of the search; determine the search space range of the particles, which must cover all possible path areas of the target waters; set the inertia weight to balance the global and local search capabilities of the particle swarm; set the individual optimal weight and the group optimal weight, which are used to measure the degree of dependence of the particle on its own historical optimal position and the global optimal position of the entire particle swarm, respectively.

[0128] According to the local risk factor distribution data, a corresponding risk factor weight is assigned to each node of the route, and the spatial coordinates of the node and the risk factor weight are used as the input of the fitness evaluation function of the particle swarm optimization algorithm:

[0129] The risk factor weight is determined by the risk factor of the spatial location of the node. The higher the risk factor weight, the greater the risk of the node's location. The spatial coordinates of the node and its corresponding risk factor weight are used as the input of the fitness evaluation function of the particle swarm optimization algorithm to provide a risk assessment basis for the optimization of each node.

[0130] This allocation method of risk factors can ensure that the particle swarm optimization algorithm focuses on avoiding high-risk areas during the path optimization process.

[0131] In each iteration, the particle swarm optimization algorithm is used to calculate the fitness value of the particle, which is evaluated based on the spatial coordinate offset of the node and the risk factor weight:

[0132] The fitness value is calculated based on the spatial coordinate offset and risk factor weight of the node. The spatial coordinate offset is used to measure the degree to which the node deviates from the ideal path, and the risk factor weight is used to reflect the risk level of the area where the node is located. The fitness value is used to evaluate whether the current state of the particle is better than the historical state, providing a basis for updating the particle speed and position.

[0133] In the process of fitness value evaluation, the deviation degree and risk factor of the path are taken into consideration at the same time to ensure that the optimization result not only meets the rationality of the path but also effectively avoids high-risk areas.

[0134] Dynamically update the speed and position of the particle, where the speed of the particle is used to control the adjustment amplitude of the node coordinates during the update process, and the current position of the particle is adjusted according to the updated speed to gradually approach the global optimal solution:

[0135] The speed of the particle determines the adjustment amplitude of the node coordinates in each iteration; the current position of the particle is adjusted according to the updated speed to gradually approach the global optimal solution; in the process of updating the speed and position, the historical optimal position of the node and the global optimal position of the particle swarm are fully considered to ensure that the optimization result can balance the capabilities of global search and local search.

[0136] After each iteration, the fitness evaluation criteria are dynamically adjusted according to the current local risk factor distribution data, so that the route nodes are gradually optimized to the low risk factor area:

[0137] When the risk factors in certain areas change, the risk factor weights are updated in real time so that the optimization algorithm can dynamically adapt to environmental changes; the fitness evaluation criteria are dynamically adjusted so that the route nodes are gradually optimized to low-risk factor areas.

[0138] When the convergence condition of the particle swarm optimization algorithm is met, the optimized route node coordinates are output as the updated route of the unmanned surface vehicle:

[0139] The convergence conditions include that the change of fitness value is less than the set threshold or the maximum number of iterations is reached; the optimized route node coordinates represent the best path for the unmanned surface vehicle to avoid the high risk factor area.

[0140] Among them, the determination of setting the threshold and the maximum number of iterations is based on the specific needs of the unmanned surface vehicle path planning and the complexity of the environment: the threshold is determined according to the mission accuracy requirements of the unmanned surface vehicle. When the change in the fitness value is less than the threshold, it indicates that the route optimization has stabilized. The stability of the path deviation and risk factor is often verified through multiple experiments, and the minimum value of the error range is taken as the threshold. The maximum number of iterations is set to prevent the algorithm from falling into local optimality or invalid iterations, combined with computing resource constraints and the scale of the target waters. Usually, experiments are performed to determine the reasonable range of iterations that can obtain a better solution in a complex environment.

[0141] The real-time position and attitude information of the unmanned surface vehicle are dynamically corrected through the Kalman filter algorithm, and the final navigation path that meets the avoidance threshold is output, including:

[0142] Get the real-time sensor data of the unmanned surface vehicle, including position coordinates, attitude angle and movement speed:

[0143] Real-time data collection of operating status is done through a variety of sensors installed on the unmanned surface vessel, including a global positioning system receiver, an inertial measurement unit, and a speed measurement module;

[0144] The position coordinates represent the current specific position of the unmanned surface vessel in space; the attitude angle describes the pitch angle, roll angle and heading angle of the unmanned surface vessel; and the movement speed represents the speed component of the unmanned surface vessel in the current running direction.

[0145] Construct a Kalman filter model, set the state vector as the real-time position and attitude information of the unmanned surface vehicle, and the observation vector as the data collected by the sensor:

[0146] According to the motion characteristics of the unmanned surface vessel and the characteristics of sensor data, a Kalman filter model is constructed: the state vector is set to the real-time position and attitude information of the unmanned surface vessel, including position coordinates, attitude angle and motion speed; the observation vector is set to the data collected by the real-time sensor, and the dimension of the observation vector is consistent with the type of sensor data.

[0147] The state transfer equation and observation equation are defined in the Kalman filter model: the state transfer equation describes the changing law of the position and posture of the unmanned surface vessel, and predicts the current state based on the state and control input at the previous moment; the observation equation describes the mapping relationship between sensor data and state vectors, and provides observation values ​​of the actual state.

[0148] Initialize the model parameters, including the initial value of the state vector, the error covariance matrix, the initial values ​​of the process noise and the observation noise, to adapt to the operating environment of the unmanned surface vehicle.

[0149] According to the real-time sensor data and the predicted value of the Kalman filter model, the error covariance matrix and Kalman gain are calculated to correct the state vector:

[0150] The state transition equation of the Kalman filter model is used to predict the state vector and error covariance matrix of the unmanned surface vehicle at the next moment; the real-time observation data collected by the sensor is compared with the model prediction value to calculate the prediction error; the error covariance matrix is ​​updated according to the prediction error to generate the optimal estimate of the current state; the Kalman gain is calculated, which is used to balance the contribution of the model prediction value and the real-time observation data to the state correction. The size of the Kalman gain is determined by the error covariance matrix and the observation noise, and the correction ratio can be adjusted dynamically.

[0151] Dynamically update the state vector and map the corrected position and attitude information to the actual running trajectory of the unmanned surface vehicle:

[0152] The state vector is corrected using the Kalman gain, and the state vector at the current moment is generated by combining the weighted result of the model prediction value and the sensor observation data; the dynamically updated state vector includes the current corrected position coordinates, attitude angle and movement speed of the unmanned surface vessel; the corrected state vector is mapped to the actual operation trajectory of the unmanned surface vessel, and the actual operation trajectory represents the state changes of the unmanned surface vessel at each moment in the form of a time series; during the dynamic update process, the sensor data is continuously collected and processed in real time to ensure the continuous update of the state vector.

[0153] Generate a navigation path that meets the avoidance threshold based on the corrected real-time position and attitude information:

[0154] According to the corrected state vector, it is judged whether the current operating state of the unmanned surface vessel meets the avoidance threshold condition; if the corrected state vector shows that the unmanned surface vessel is approaching a high-risk area, the navigation path is adjusted according to the path planning algorithm; the navigation path that meets the avoidance threshold is output, and the path is dynamically generated based on the corrected position and attitude information to ensure that the unmanned surface vessel can avoid the high-risk area; the generated navigation path is transmitted to the control module of the unmanned surface vessel in real time as an operating instruction for the next stage.

[0155] Example 2: The difference between Example 2 of the present invention and Example 1 is that this example introduces an unmanned surface vessel path planning system.

[0156] Figure 2 A structural schematic diagram of an unmanned surface vessel path planning system of the present invention is given, which includes a data acquisition and fusion module, an energy identification and analysis module, a path deviation evaluation module, a route optimization planning module and a dynamic correction path module.

[0157] Data acquisition and fusion module: obtains the fluctuation data of the target water area, and performs multi-source sensor fusion processing on the fluctuation data to generate basic data for identifying local wave interference areas.

[0158] Energy identification and analysis module: uses a deep neural network algorithm to identify energy concentration areas based on basic data and outputs wave interference intensity distribution.

[0159] Path deviation assessment module: Path deviation analysis is performed on the historical navigation trajectory of the unmanned surface vessel based on the wave interference intensity distribution, local risk factors are generated and avoidance thresholds are set.

[0160] Route optimization planning module: The particle swarm optimization algorithm is used to iteratively plan the preset route of the unmanned surface vessel, and the updated node coordinates are adjusted according to the local risk factors during each iteration.

[0161] Dynamic correction path module: The real-time position and attitude information of the unmanned surface vessel are dynamically corrected through the Kalman filter algorithm, and the final navigation path that meets the avoidance threshold is output.

[0162] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0163] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0164] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0165] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0166] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0167] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0168] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0169] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0170] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0171] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A path planning method for an unmanned surface vessel, characterized in that: The steps include: S1: Obtain the wave data of the target water area, and perform multi-source sensor fusion processing on the wave data to generate basic data for identifying the local wave interference area; S2: Use deep neural network algorithm to identify energy concentration areas on basic data and output wave interference intensity distribution; S3: Analyze the path deviation of the historical navigation trajectory of the unmanned surface vessel based on the wave interference intensity distribution, generate local risk factors and set avoidance thresholds; S4: Use the particle swarm optimization algorithm to iteratively plan the preset route of the unmanned surface vehicle, and adjust the updated node coordinates according to the local risk factor during each iteration; S5: The real-time position and attitude information of the unmanned surface vehicle is dynamically corrected through the Kalman filter algorithm, and the final navigation path that meets the avoidance threshold is output.

2. The unmanned surface vehicle path planning method according to claim 1, characterized in that: Obtain the wave data of the target waters, and perform multi-source sensor fusion processing on the wave data to generate basic data for identifying the local wave interference area, including: Use multiple sensors installed on the unmanned surface vessel to obtain fluctuation data of the target water area, including radar sensors, sonar sensors and laser rangefinders; Temporally and spatially align the water surface wave height data obtained by the radar sensor, the underwater wave amplitude data obtained by the sonar sensor, and the water surface texture data obtained by the laser rangefinder; Based on the weighted average fusion algorithm, multi-source sensor fusion processing is performed on the water surface wave height data, underwater wave amplitude data and water surface texture data after time and space alignment to generate fusion data for characterizing local wave characteristics; Information including the wave frequency, wave height change rate and wave direction is extracted from the fused data to generate basic data for identifying local wave interference areas.

3. The unmanned surface vehicle path planning method according to claim 1, characterized in that: The deep neural network algorithm is used to identify the energy concentration area of ​​the basic data and output the wave interference intensity distribution, including: The basic data are used as input data of the deep neural network model. The basic data include the distribution of fluctuation frequency, the distribution of wave height change trend and the distribution of fluctuation propagation direction. Construct a multi-layer deep neural network model, including an input layer, multiple hidden layers and an output layer, wherein the input layer is used to receive the multi-dimensional features of the basic data, the hidden layer includes multiple fully connected layers and activation functions, the activation function uses a rectified linear unit, and the output layer is used to generate the spatial distribution probability of the energy concentration area; The deep neural network model is trained with the training data including the wave characteristic data based on the actual sea area and the corresponding energy concentration area label data; The basic data to be analyzed is input into the trained deep neural network model, and the energy concentration area distribution probability map of the target water area is obtained through model calculation; The wave interference intensity is calculated according to the energy concentration area distribution probability map to generate wave interference intensity distribution data, which includes the wave interference intensity value of each spatial point.

4. The unmanned surface vehicle path planning method according to claim 3, characterized in that: The wave interference intensity is calculated based on the energy concentration area distribution probability map to generate the wave interference intensity distribution data, specifically: According to the probability map of energy concentration area distribution, the wave interference intensity value of each spatial point is calculated: I = P·F; where I represents the wave interference intensity, P represents the energy concentration probability value, and F represents the wave frequency of the corresponding spatial point; Generation of wave interference intensity distribution data: Filter the wave interference intensity values ​​to remove possible noise interference in the probability map; integrate the filtered wave interference intensity values ​​into distribution data to form the wave interference intensity distribution of each spatial point in the target waters.

5. The unmanned surface vehicle path planning method according to claim 1, characterized in that: Based on the wave interference intensity distribution, the path deviation analysis of the historical navigation trajectory of the unmanned surface vessel is carried out to generate local risk factors and set avoidance thresholds, including: The wave interference intensity distribution data is superimposed and analyzed with the historical navigation trajectory of the unmanned surface vessel, wherein the historical navigation trajectory data includes the spatial coordinates and time series information of the navigation path; Calculate a deviation vector based on the actual path and the planned path of the historical navigation trajectory of the unmanned surface vessel; Based on the wave interference intensity value of each spatial point in the wave interference intensity distribution data, the weight of the deviation vector is corrected to obtain a path deviation correction value; Map the path deviation correction value to the spatial point of the local water area to generate the local risk factor corresponding to each spatial point. The local risk factor is positively correlated with the path deviation correction value. The avoidance threshold is set according to the distribution characteristics of the local risk factor. The avoidance threshold is the condition that the route needs to be adjusted when the local risk factor is higher than the preset value; The local risk factors are normalized to form normalized risk factor distribution data.

6. The unmanned surface vehicle path planning method according to claim 1, characterized in that: The particle swarm optimization algorithm is used to iteratively plan the preset route of the unmanned surface vehicle, and the updated node coordinates are adjusted according to the local risk factors during each iteration, including: The preset route of the unmanned surface vehicle is discretized into multiple nodes, each node contains the corresponding spatial coordinate information, and the node is used as the initial solution of the particle swarm optimization algorithm; Initialize the relevant parameters of the particle swarm optimization algorithm, including the number of particles, particle speed, particle search space range, inertia weight, individual optimal weight and group optimal weight; According to the local risk factor distribution data, a corresponding risk factor weight is assigned to each node of the route, and the spatial coordinates of the node and the risk factor weight are used as the input of the fitness evaluation function of the particle swarm optimization algorithm; In each iteration, the particle swarm optimization algorithm is used to calculate the fitness value of the particle, which is evaluated based on the spatial coordinate offset of the node and the risk factor weight; Dynamically update the speed and position of particles. The speed of particles is used to control the adjustment amplitude of node coordinates during the update process, and the current position of particles is adjusted according to the updated speed to gradually approach the global optimal solution. After each iteration, the fitness evaluation criteria are dynamically adjusted according to the current local risk factor distribution data, so that the route nodes are gradually optimized to the low risk factor area; When the convergence condition of the particle swarm optimization algorithm is met, the optimized route node coordinates are output as the updated route of the unmanned surface vehicle.

7. The unmanned surface vehicle path planning method according to claim 1, characterized in that: The real-time position and attitude information of the unmanned surface vehicle are dynamically corrected through the Kalman filter algorithm, and the final navigation path that meets the avoidance threshold is output, including: Acquire real-time sensor data of the unmanned surface vehicle, including position coordinates, attitude angle and movement speed; Construct a Kalman filter model, set the state vector as the real-time position and attitude information of the unmanned surface vehicle, and the observation vector as the data collected by the sensor; According to the real-time sensor data and the predicted value of the Kalman filter model, the error covariance matrix and Kalman gain are calculated to correct the state vector; Dynamically update the state vector and map the corrected position and attitude information to the actual running trajectory of the unmanned surface vessel; A navigation path that meets the avoidance threshold is generated based on the corrected real-time position and attitude information.

8. An unmanned surface vessel path planning system, used to implement an unmanned surface vessel path planning method according to any one of claims 1 to 7, characterized in that: It includes data acquisition and fusion module, energy identification and analysis module, path deviation assessment module, route optimization planning module and dynamic correction path module; Data acquisition and fusion module: obtains the wave data of the target water area, and performs multi-source sensor fusion processing on the wave data to generate basic data for identifying the local wave interference area; Energy identification and analysis module: uses deep neural network algorithm to identify energy concentration areas based on basic data and outputs wave interference intensity distribution; Path deviation assessment module: Analyze the path deviation of the historical navigation trajectory of the unmanned surface vessel based on the wave interference intensity distribution, generate local risk factors and set avoidance thresholds; Route optimization planning module: It uses the particle swarm optimization algorithm to iteratively plan the preset route of the unmanned surface vehicle, and adjusts the updated node coordinates according to the local risk factor during each iteration; Dynamic correction path module: The real-time position and attitude information of the unmanned surface vessel are dynamically corrected through the Kalman filter algorithm, and the final navigation path that meets the avoidance threshold is output.

Images