UUV coverage path energy efficiency optimization method and offshore surveying and mapping system

Abstract

The invention relates to a UUV coverage path energy efficiency optimization method and an offshore surveying and mapping system, and the method comprises the following steps: obtaining a satellite remote sensing image, multi-beam sounding data and underwater topographic data of an offshore region, recognizing a sea-land boundary contour, and constructing an offshore seabed three-dimensional surveying and mapping region model; a UUV energy consumption model is constructed, a surveying and mapping area coverage rate maximization and total energy consumption minimization are taken as targets, a surveying and mapping area model is combined, an intelligent optimization algorithm is adopted to carry out global path planning, and a UUV surveying and mapping operation path is determined; the underwater environment is monitored in real time through a sensor carried by the UUV, whether an original planned path is feasible or not is judged, and when the original planned path is not feasible, local path re-planning is carried out by taking minimization of extra energy consumption as a target and adopting an improved path search algorithm. Compared with the prior art, the method has the effects of improving the accuracy and integrity of a surveying and mapping area, reducing the energy consumption of the UUV, improving the surveying and mapping efficiency and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned underwater vehicle, in particular to a UUV coverage path energy efficiency optimization method and an offshore mapping system. BACKGROUND

[0002] The offshore area is a key area in the fields of marine resource development, port construction, coastal engineering and marine ecological environment protection, and accurate mapping of the seabed topography of the offshore area is of great significance. Unmanned underwater vehicle (UUV) has become an important tool for mapping the seabed of the offshore area due to its characteristics of autonomous operation, high flexibility and adaptability to complex underwater environment.

[0003] However, there are many technical difficulties in using UUV for nearshore seabed mapping. First, the environment of the nearshore area is extremely complex, and the shape of the sea-land boundary is variable, affected by tides, waves, river estuaries and other factors. Traditional boundary identification methods cannot accurately outline the sea-land boundary profile. The existing mapping area modeling method is often based on simple geometric models or limited prior data, which cannot truly reflect the complex topographic features of the nearshore seabed, which leads to the UUV in the execution of the mapping task, easy to appear mapping area omission or repeated mapping situation, seriously affect the mapping efficiency and accuracy. Second, the energy supply of UUV is limited, and the mapping task usually needs long-time continuous operation. Under the energy constraint, how to plan the mapping operation coverage path of UUV so as to complete the mapping task while reducing energy consumption as much as possible and prolonging the operation time is a problem to be solved. The existing path planning method does not fully consider the energy factor, or only uses a simple energy consumption model, which leads to the UUV in the actual operation, may be unable to complete the task due to energy depletion, or the energy utilization efficiency is low, causing energy waste. Chinese patent CN115655274A discloses a double-layer planning framework-based energy consumption optimal underwater area coverage method under the influence of sea current, comprising: processing the known to-be-operated sea area map to construct an environment map; using the coverage path and energy consumption index as decision variables to express the problem as a double-layer planning model; the upper-layer algorithm uses a biological excitation neural network algorithm to solve the coverage path and transmits it to the lower-layer algorithm; the lower-layer uses an ant colony optimization algorithm to solve the energy consumption problem, taking the sea current as a constraint to calculate the energy consumption fitness value and return it to the upper-layer algorithm, if there is no feasible path, find an escape path and return it to the upper-layer algorithm; the upper-layer algorithm makes path decision based on the results of the lower-layer algorithm, and iterates until the task area coverage is completed, outputting the optimal coverage path and AUV propulsion speed. However, in the mapping operation process, due to the uncertainty of the underwater environment, such as sudden changes in water flow, obstacles, etc., the original mapping path may no longer be applicable, and this method does not consider this. The current local re-planning technology lacks effective energy-saving optimization strategies, and often increases unnecessary energy consumption when adjusting the path, further exacerbating the tight energy supply situation of UUV.

[0004] In summary, the existing UUV nearshore seabed mapping technology has obvious deficiencies in sea-land boundary profile identification and mapping area modeling, coverage path planning under energy constraint, and local re-planning of mapping operation, and an efficient and energy-saving UUV coverage path energy efficiency optimization method is urgently needed to meet the growing demand for nearshore seabed mapping. SUMMARY

[0005] The purpose of the present application is to provide a UUV coverage path energy efficiency optimization method and an offshore mapping system, which breaks through key technologies such as sea-land boundary contour recognition based on deep learning, mapping area modeling, UUV mapping operation coverage planning under energy constraints, and local mapping operation re-planning based on energy-saving optimization model, to ensure efficient mapping of UUV while reducing UUV path energy consumption.

[0006] The purpose of the present application can be achieved by the following technical solutions: A UUV coverage path energy efficiency optimization method for offshore mapping, comprising the following steps: Data acquisition and preprocessing: acquiring satellite remote sensing images, multi-beam sounding data and underwater topographic data of the offshore area; performing radiation correction and geometric correction on the satellite remote sensing images, and performing noise removal and data interpolation processing on the multi-beam sounding data; Sea-land boundary contour recognition and mapping area modeling based on deep learning: inputting the preprocessed satellite remote sensing images into the trained deep convolutional neural network to recognize the sea-land boundary contour; combining the sea-land boundary contour with the preprocessed multi-beam sounding data and underwater topographic data to construct a three-dimensional mapping area model of the offshore seabed; UUV mapping operation coverage planning under energy constraints: constructing a UUV energy consumption model, comprehensively considering the influence of navigation, attitude adjustment and sensor operation on energy consumption, taking the maximum mapping area coverage rate and the minimum total energy consumption as the target, combining the mapping area model, and using intelligent optimization algorithm for global path planning to determine the UUV mapping operation path; Local re-planning of mapping operation based on energy-saving optimization model: real-time monitoring of underwater environment by the sensor carried by the UUV, obtaining obstacle information and flow velocity data and preprocessing; judging whether the original planned path is feasible based on the obstacle information and flow velocity, and when the original planned path is not feasible, taking the minimization of additional energy consumption as the target, using an improved path search algorithm to perform local path re-planning under the constraint conditions of time, path and UUV performance.

[0007] The deep convolutional neural network adopts a U-Net network structure, including an encoding layer and a decoding layer, the encoding layer extracts image features through convolution and maximum pooling operations on the input satellite remote sensing images, the decoding layer restores image resolution through deconvolution operations and performs feature splicing, outputs a sea-land boundary probability map, and obtains a sea-land boundary contour through threshold segmentation.

[0008] The UUV energy consumption model is represented as: Wherein, is the total energy consumption of the UUV, is the navigation energy consumption of the UUV in the first segment path,n The number of segments for the discretized path. Energy consumption for UUV attitude adjustment For sensor operating power consumption, The constant energy consumption per unit time for sensor operation. Total operation time; Assuming UUV is in the first If a segment of the path is traveled at a constant speed, then the energy consumption of that segment of the journey is... for: in, For the density of water, The underwater drag coefficient, For the cross-sectional area of ​​the UUV, For UUV in the A constant sailing speed along the route. For UUV in the The sailing distance of a segment of the route; The energy consumption for UUV attitude adjustment is expressed as: in, Energy consumption coefficient for attitude adjustment Adjust the angle to maintain the posture. To adjust the frequency.

[0009] The specific method of using intelligent optimization algorithms for global path planning is as follows: based on the coverage rate of the surveyed area... Maximize and total energy consumption Minimize to a bi-objective function: The bi-objective function is transformed into a single-objective function: in, For the theoretical maximum coverage, The maximum energy that a fully charged UUV battery can provide. , These are the weighting coefficients; Using the single objective function as the optimization objective, a non-dominated sorting genetic algorithm is employed to solve the problem, obtaining the optimal path scheme under the conditions of satisfying energy consumption constraints, path constraints, speed constraints, and depth constraints.

[0010] The process involves using sensors mounted on the UUV to monitor the underwater environment in real time, acquiring obstacle information and water flow velocity data, and then preprocessing the data. Obstacle information is acquired using a multi-beam sonar sensor mounted on the UUV, and the detection data from the multi-beam sonar sensor is represented as a point cloud set. ,in For points in a point cloud, These are the three-dimensional coordinates of the point. The point cloud count; water flow velocity is obtained using an acoustic Doppler current profiler mounted on the UUV, at time... The measured water flow velocity vector is ,in, These represent the water flow at... , , The velocity component in the direction; For point cloud data acquired by multibeam sonar sensors, a statistical filtering algorithm is used to remove outliers: assuming a certain point in the point cloud data... To its The average distance between nearest neighbors is ,like Greater than the global average distance With preset multiple If the product of the product of the product of the product and the product of the product, then the point is determined to be an outlier and is removed. The water flow velocity data collected by the acoustic Doppler current profiler is processed using a Kalman filter algorithm to predict the water flow velocity at the next moment.

[0011] The minimized additional energy consumption is expressed as: min in, For additional energy consumption, Increase in navigation energy consumption To account for the increase in energy consumption for attitude adjustment, let the original planned path be... The replanned path is ,but in, , These represent the number of path segments before and after the replanning. , The first and second phases of the replanning were respectively The location coordinates of the path segment. For the density of water, The underwater drag coefficient, For the cross-sectional area of ​​the UUV, The velocity of the UUV relative to the ground. , For the UUV's own travel speed, For water flow velocity, For the first Travel time for a segment of the route , This is the length of the path segment. For the replanned path of UUV The magnitude of the velocity relative to the ground at any given moment. This represents the speed of the UUV relative to the ground at the corresponding location and time along its original planned path. Energy consumption coefficient for attitude adjustment For the first The angle of the path turning point in this adjustment. To adjust the number of times.

[0012] The constraints on time, path, and UUV performance are expressed as follows: Time constraint: The replanned path must be completed within a specified time. The surveying and mapping task was completed within the specified timeframe, namely: in, For the replanned path of UUV The magnitude of the velocity relative to the ground at any given moment. For the first Length of the path segment The number of path segments after replanning; Path constraints: The replanned path must be within the survey area model and avoid obstacle areas; UUV performance constraints: The speed, depth, and attitude adjustment angle of the UUV must meet the corresponding upper and lower limits.

[0013] The improved path search algorithm employs an improved fast search random tree algorithm, introducing a heuristic function to guide the search of sampling points. During tree expansion, it reselects parent nodes to connect new nodes to the parent node with the lowest cost, thereby reducing the overall path cost. The heuristic function... Based on the current node To the target point Euclidean distance and water flow influencing factors Construct, represented as: in, The value is dynamically adjusted based on the water flow velocity; the greater the water flow velocity, the larger the value.

[0014] A near-coastal mapping system, implemented based on the method described above, includes: The data processing module is used to perform the data acquisition and preprocessing steps and the deep learning-based land-sea boundary contour recognition and mapping area modeling steps. The path planning module is used to execute the UUV mapping operation coverage planning step based on energy constraints and the mapping operation local replanning step based on the energy-saving optimization model. The operation monitoring module is used to display the UUV's location, navigation path, energy consumption, and mission progress information in real time. It displays the UUV's operation trajectory in the survey area in the form of a 3D map through a visual interface. Based on the UUV's real-time energy consumption data and remaining path, it uses a gray prediction model to predict energy consumption. According to the energy consumption prediction results, when the UUV's remaining energy is lower than a preset threshold, the system automatically issues an early warning and adjusts the path or terminates the mission according to preset strategies.

[0015] The system also includes an intelligent decision-making and optimization module, which employs a reinforcement learning algorithm and sets the reward function as follows: in, This is the standard task completion time. This refers to the actual time it took to complete the task. For the mission's budgeted energy consumption, For UUV energy consumption, This is a value used to assess the accuracy of the surveying and mapping. , , These are the weighting coefficients; By continuously adjusting the path planning parameters and UUV operation parameters, the reward function value is maximized. This enables continuous optimization of surveying and mapping operations.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention overcomes the problem of poor adaptability of traditional methods in nearshore areas by integrating multi-source data (satellite images, depth sounding data) and using deep learning to intelligently identify complex and ever-changing land-sea boundaries. The constructed three-dimensional mapping area model more realistically reflects the seabed topography, fundamentally avoiding mapping blind spots or repeated mapping caused by inaccurate models, and ensuring the integrity and reliability of mapping data.

[0017] (2) This invention embeds an accurate UUV energy consumption model into global path planning, with the dual optimization objectives of the highest coverage and the lowest total energy consumption, so that the UUV path is no longer a simple geometric coverage, but an energy-optimal coverage, thereby significantly extending the effective working time or completing a larger range of mapping tasks under a limited energy budget.

[0018] (3) The present invention introduces a local replanning mechanism based on real-time monitoring data. When encountering unexpected obstacles or strong water flow, the UUV does not simply stop the task or force its way through, but can dynamically and intelligently adjust the path. This not only ensures the continuity and safety of the surveying task, but also ensures that the local adjustment will not have too much impact on the overall energy budget.

[0019] (4) This invention fully considers the actual challenges of nearshore operations (complex environment, limited energy, high uncertainty). Through the synergistic strategy of global planning and local replanning, it has both macro-optimality and flexibility to cope with micro-dynamic changes. It can effectively cope with various emergencies in the real marine environment and has strong engineering application potential and robustness. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system structure diagram of the present invention. Detailed Implementation

[0021] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0022] This embodiment discloses a UUV coverage path energy efficiency optimization method for nearshore seabed mapping and a nearshore mapping system based on this method. Addressing the difficulties in identifying land-sea boundaries and low energy efficiency faced by UUVs in nearshore seabed mapping, the method first collects and preprocesses satellite remote sensing images and multibeam bathymetry data to construct a deep convolutional neural network to identify land-sea boundary contours. A three-dimensional mapping area model of the nearshore seabed is then established using multi-source data. Next, a UUV energy consumption model is constructed, aiming for full coverage of the mapping area and minimizing energy consumption. An intelligent optimization algorithm is used for global path planning, while sensors onboard the UUV monitor the underwater environment in real time. When the original path is infeasible, local path replanning is performed with the goal of minimizing additional energy consumption. Finally, a UUV nearshore mapping cabling software with a layered architecture integrates the above technologies, achieving full-process management from data processing and path planning to operation monitoring and optimization, forming a nearshore mapping system. This invention improves the accuracy and completeness of the mapping area, reduces UUV energy consumption, extends operation time, and improves mapping efficiency, showing promising prospects in marine resource development and port construction.

[0023] This embodiment first provides a method for optimizing the energy efficiency of UUV coverage paths in near-coastal mapping, such as... Figure 1 As shown, it includes the following steps: S1, Data Acquisition and Preprocessing: Acquire satellite remote sensing images, multibeam bathymetry data, and underwater topographic data of the nearshore area; perform radiometric and geometric correction on the satellite remote sensing images, and perform noise removal and data interpolation on the multibeam bathymetry data.

[0024] Satellite remote sensing images covering visible light, infrared, and other wavelengths are acquired through satellite remote sensing. Multibeam echo sounding system is used to collect multibeam echo sounding data. And underwater topographic data, with the sampling interval set to Meters, ensuring that the data density meets the modeling requirements, among which, Image pixel coordinates, x The x-axis of the image, y The vertical axis of the image, The coordinates of the depth sounding data sampling points, s This represents the distance coordinates of the sampling point along the flight path. t This represents the distance coordinates of the sampling point across the course of the route. Additionally, historical ocean environmental data, including information on tides and current speeds, can be obtained.

[0025] Satellite remote sensing images Radiation correction is performed to eliminate the influence of factors such as light: in, This is the gain coefficient. The bias coefficient is determined through statistical analysis of historical calibration data. The image is after radiometric correction; Then, geometric correction is performed, selecting at least [number missing] areas in the nearshore region. One ground control point, using the ground control point With image coordinates Polynomial transformation relationships correct image geometric distortions: , in, The degree of the polynomial is selected based on the degree of image distortion, and is generally taken as [degree]. ; , The coefficients of the polynomial transformation are... x-axis of the image x of i Power of 1 The vertical axis of the image y of j Power of 1.

[0026] Multibeam bathymetry data Noise removal is achieved using a median filtering algorithm: in, The radius of the filtering window is determined based on the data noise level; in this embodiment, it is set to 2.

[0027] Then, the noise-removed multibeam bathymetry data is transformed from the original data coordinate system of the multibeam bathymetry system to a unified geodetic coordinate system, and the discrete bathymetry data is interpolated into regular three-dimensional terrain data using the kriging interpolation method: in, For regular grid points in coordinates Interpolation water depth at the location, For the first i The weighting coefficients for each sampling point are determined by solving the semivariogram. The number of neighboring sampling points participating in this interpolation calculation. For neighboring sampling points i coordinates The depth measurement at that location.

[0028] S2, Deep Learning-Based Land-Sea Boundary Contour Recognition and Mapping Area Modeling: Preprocessed satellite remote sensing images are input into a trained deep convolutional neural network to identify land-sea boundary contours; combined with land-sea boundary contours and preprocessed multibeam bathymetry data and underwater topographic data, a three-dimensional mapping area model of the nearshore seabed is constructed.

[0029] S21, Land-sea boundary contour recognition In this embodiment, the deep convolutional neural network adopts the U-Net network structure, which includes an encoding layer and a decoding layer. The encoding layer extracts image features from the input satellite remote sensing image through convolution and max pooling operations. The decoding layer restores the image resolution and performs feature stitching through deconvolution operations, outputs a land-sea boundary probability map, and obtains the land-sea boundary contour through threshold segmentation.

[0030] The encoding layer extracts image features through convolution operations. The convolution operation formula is as follows: in, For the convolutional feature map in The value at that location, The input image contains the pixel value at the corresponding location. For the convolution kernel in The value at that location, The kernel size; Subsequently, max pooling is used to reduce data dimensionality. in, This is the pooling step size.

[0031] The decoding layer restores the image resolution through deconvolution. The deconvolution formula is: Finally, the feature maps of the corresponding scale of the encoding layer are concatenated and fused with the feature maps of the decoding layer.

[0032] In one embodiment, the deep convolutional neural network is configured with an encoding layer comprising: There are one downsampling module, each module consists of... Each convolutional layer and It consists of a maximum pooling layer; the decoding layer contains... Each upsampling module consists of [number] modules. One deconvolutional layer and It consists of several convolutional layers, and features are stitched together at the corresponding scale.

[0033] The network is trained using the cross-entropy loss function, and the network parameters are updated using the Adam optimizer. in, for Time-model parameters, For learning rate, For first-order moment estimation, For second-order moment estimation, To prevent extremely small constants with a denominator of zero.

[0034] By having professionals annotate the land and sea boundaries of the collected satellite remote sensing images, a system is generated that includes... A dataset of labeled images, with training, validation, and test sets divided in a 7:1:2 ratio. Training parameters: cross-entropy loss function, Adam optimizer learning rate. First-order moment estimation Second-order moment estimation , The training batch size is set to Training cycle Then, the training set images are input into the network, the network parameters are updated through the backpropagation algorithm, the model performance is monitored on the validation set, training is stopped when the validation set loss no longer decreases, and the accuracy of the model in identifying land and sea boundaries is evaluated using the test set.

[0035] After training, the preprocessed satellite remote sensing images are input into a deep convolutional neural network, which outputs a probability map of the land-sea boundary. The land-sea boundary contour is obtained through threshold segmentation: in, The segmentation threshold is adaptively determined using the Otsu algorithm.

[0036] S22, Construction of Survey Area Model The extracted land-sea boundary contours are combined with processed multibeam bathymetry data and underwater topographic data, using a triangular mesh modeling method. For the point set in three-dimensional space... ,in, For point The three-dimensional coordinates Given the number of points in the point set, a triangular mesh is constructed using the Delaunay triangulation algorithm, which satisfies the maximum-minimum angle criterion, meaning that the minimum interior angle of each triangle is maximized among all possible triangulations. In this embodiment, the minimum interior angle threshold is set to [value missing]. The aspect ratio threshold is The quality of the triangular mesh is assessed by minimizing the interior angle. Measured in terms of aspect ratio: in, It is the longest side of the triangle. It is the shortest side.

[0037] When the preset quality threshold is not met, optimization operations such as edge swapping and vertex insertion are employed. Ultimately, a three-dimensional mapping area model that accurately reflects the topographic features of the nearshore seabed is constructed, providing an accurate geographic information foundation for UUV path planning.

[0038] S3, UUV mapping operation coverage planning under energy constraints: Construct a UUV energy consumption model, comprehensively consider the impact of navigation, attitude adjustment and sensor operation on energy consumption, with the goal of maximizing the coverage of the mapping area and minimizing the total energy consumption, and combine the mapping area model to use intelligent optimization algorithms for global path planning to determine the UUV mapping operation path.

[0039] In nearshore seabed mapping scenarios, UUVs are limited by battery capacity, and the complex nearshore currents and topography make it difficult for traditional path planning to balance mapping coverage and energy consumption. This invention constructs a UUV energy consumption model, combined with a mapping area model, with the goal of minimizing energy consumption and maximizing mapping coverage, and achieves global path planning through an intelligent optimization algorithm.

[0040] S31, Construction of UUV Energy Consumption Model The energy consumption of a UUV mainly comes from navigation, attitude adjustment, and sensor operation, with navigation energy consumption accounting for the largest proportion. and sailing distance ,speed ,depth underwater drag coefficient UUV cross-sectional area and water density The relevant calculation formula is as follows: Since the speed and depth of a UUV may vary in actual operations, they are discretized. It is assumed that the UUV is at the... The path segment travels at a constant speed ,depth sailing distance The energy consumption of this segment of the voyage for: depth It doesn't appear directly in the two formulas above, but it works by affecting the density of water. and drag coefficient Indirectly affecting energy consumption. In this embodiment, it is assumed that the first... i Segment depth h Constant, therefore and Treating it as a constant is suitable for scenarios where depth changes are small. If the depth changes significantly, it may be necessary to... and Modeling as depth h The function.

[0041] Attitude adjustment energy consumption With attitude adjustment angle Adjusting the frequency Related can be represented as: in, The energy consumption coefficient for attitude adjustment was determined through UUV dynamic experiments.

[0042] Sensor operating power consumption It remains essentially constant within a unit of time, let it be denoted as . .

[0043] Total energy consumption for: in, The total energy consumption of UUV n The number of segments for the discretized path. The constant energy consumption per unit time for sensor operation. This represents the total operation time.

[0044] In this embodiment, different sailing speeds were measured through UUV pool experiments and actual sea trials. ,depth Posture adjustment angle Based on the energy consumption data, determine the underwater drag coefficient. UUV cross-sectional area square meters, attitude adjustment energy consumption coefficient Energy consumption per unit time of sensor operation watt.

[0045] S32, Objective function and constraints of global path planning The objective function is the coverage of the surveyed area. Maximize and total energy consumption Minimize to a bi-objective function: Transform the bi-objective function into a single-objective function: in, The theoretical maximum coverage (generally 100%) ), The maximum energy that a fully charged UUV battery can provide. , These are the weighting coefficients. Adjust according to actual needs; for example, if energy consumption is a concern, the power can be increased. .

[0046] The constraints include: 1) Energy consumption constraints: This ensures that the UUV will not be interrupted due to energy depletion during the mission.

[0047] 2) Path Constraints: The UUV path must remain within the 3D mapping area model constructed based on deep learning and avoid obstacles. Let the 3D mapping area model be... The obstacle area is Then the path point Must meet and .

[0048] 3) Depth constraint: UUV travel speed Must meet ,in, , These represent the minimum and maximum speeds of a UUV, and the depth. Must meet ,in, , (These refer to the safe operating depth range for UUVs).

[0049] In this embodiment, the weight coefficients in the objective function are set. , , , The battery capacity of the UUV is determined to be 1000 watt-hours. Constraint: Define the minimum travel speed of the UUV. meters per second, maximum sailing speed meters per second; minimum safe operating depth meters, maximum depth Meters; minimum angle for attitude adjustment , the maximum angle .

[0050] S33, Solved by intelligent optimization algorithm Using a single objective function as the optimization objective, the non-dominated sorting genetic algorithm (NSGA-II) is employed to solve the problem, obtaining the optimal path scheme under the conditions of satisfying energy consumption constraints, path constraints, speed constraints, and depth constraints.

[0051] The algorithm searches for Pareto front solutions in the solution space by using fast non-dominated sorting and crowding calculation.

[0052] Encoding method: The mapping path of UUV is encoded as a chromosome, and the chromosome consists of a series of path point coordinates. composition.

[0053] Fitness calculation: based on the objective function Calculate the fitness of each chromosome; the higher the fitness, the better the path scheme.

[0054] Selection operation: A tournament selection method is used to randomly select from the population. Individuals ( For the scale of a tournament, it is generally taken as The individual with the best fitness is selected to enter the next generation.

[0055] Crossover operation: The Partial Matching Crossover (PMX) algorithm is used to randomly select two crossover points in the two parent chromosomes, exchange gene segments between the crossover points, and ensure the legitimacy of the chromosomes through conflict resolution strategies.

[0056] Mutation operation: Making random, small-scale adjustments to genes (path point coordinates) in a chromosome, such as... Become To increase population diversity, among which, , , It is a random, tiny offset.

[0057] Through multiple generations of evolution, the algorithm converges to obtain a UUV mapping operation path scheme that satisfies the constraints and optimizes the objective function, thus realizing coverage planning under energy constraints.

[0058] In this embodiment, the population size is set to 100, the number of iterations is 100, the crossover probability is 0.8, and the mutation probability is 0.05. The 3D model of the survey area is imported into the algorithm, and the coordinates of the path points are used as chromosome codes. Through operations such as selection, crossover, and mutation, the algorithm searches for the UUV surveying operation path scheme that optimizes the objective function under the constraints.

[0059] S4, Local replanning of surveying operations based on energy-saving optimization model: The underwater environment is monitored in real time by the sensors on the UUV, and obstacle information and water flow velocity data are obtained and preprocessed; the feasibility of the original planned path is determined based on the obstacle information and water flow velocity. When the original planned path is not feasible, an improved path search algorithm is adopted with the goal of minimizing additional energy consumption, and local path replanning is performed under the constraints of time, path and UUV performance.

[0060] In nearshore seabed mapping operations, UUVs often encounter sudden currents and unexpected obstacles, rendering the originally planned route unusable. Traditional local replanning methods often neglect energy consumption optimization or only employ simple path adjustment strategies, failing to meet the needs of UUVs to efficiently complete mapping tasks under energy-constrained conditions. This invention proposes a local replanning method for mapping operations based on an energy-saving optimization model. Combining the characteristics of the nearshore seabed environment, it achieves low-energy path replanning through real-time environmental monitoring, the construction of an energy-saving optimization model, and the use of an efficient path search algorithm.

[0061] S41, Real-time Environmental Monitoring and Data Processing Obstacle information is acquired using a 500kHz multibeam sonar sensor mounted on the UUV, with a detection range of [missing information]. The detection data from the multibeam sonar sensor is represented as a point cloud set. ,in For points in a point cloud, These are the three-dimensional coordinates of the point. This represents the number of point clouds.

[0062] Water flow velocity is obtained using an acoustic Doppler current profiler (ADCP) mounted on the UUV, at time... The measured water flow velocity vector is ,in, These represent the water flow at... , , The velocity component in the direction.

[0063] For point cloud data acquired by multibeam sonar sensors, a statistical filtering algorithm is used to remove outliers: assuming a certain point in the point cloud data... To its The average distance between nearest neighbors is ,like Greater than the global average distance With preset multiple If the product of the product of the two values ​​is used, the point is determined to be an outlier and is removed. In this embodiment, Set to 10, global average distance multiple Remove outliers.

[0064] The water velocity data acquired by the acoustic Doppler current profiler is processed using a Kalman filter algorithm to predict the water velocity at the next moment. The state equation for the Kalman filter is: The observation equation is ,in It is a state vector (containing information such as water flow velocity). Here is the state transition matrix. For the control matrix, For control vectors, For process noise, For the observation vector, For the observation matrix, To reduce noise, Kalman filtering can be used to obtain more accurate predicted water flow velocity values. .

[0065] S42, Determine if the original planned path is feasible. First, based on obstacle information, if any segment of the original path intersects with the area where the obstacle is located, the path is determined to be infeasible because there is a risk of collision.

[0066] Secondly, for each segment of the original path, the UUV needs to maintain a ground velocity vector to travel as planned. The UUV's water velocity vector is calculated based on the water flow velocity and the ground velocity vector, and the required thrust is then calculated accordingly. It is then checked whether the required thrust exceeds the UUV's maximum thrust. If it does, it means that even with full thrust, the UUV cannot counteract the water flow and cannot follow the original path; therefore, the original path is not feasible.

[0067] The original path can only be followed if both obstacle information and water flow speed indicate that the original path is feasible; otherwise, local replanning is required.

[0068] S43, Energy-saving optimization model construction When constructing the energy-saving optimization model, the goal is to minimize the additional energy consumption after path replanning. The target is to reduce energy consumption. Additional energy consumption mainly includes increased navigation energy consumption due to path changes and energy consumption for attitude adjustment.

[0069] Let the original planned route be The replanned path is , , These represent the number of path segments before and after the replanning. , The first and second phases of the replanning were respectively The location coordinates of the segment path.

[0070] Increase in navigation energy consumption The calculations consider the impact of water flow on UUV navigation. Under the influence of water flow, the speed of the UUV relative to the ground... With the speed of the UUV itself and water flow velocity The relationship is Then the first Travel time of segment path for: in, This is the length of the path segment. Therefore, the increase in navigation energy consumption... for: in, For the density of water, The underwater drag coefficient, For the cross-sectional area of ​​the UUV, For the replanned path of UUV The magnitude of the velocity relative to the ground at any given moment. This represents the speed of the UUV relative to the ground at the corresponding location and time along its original planned path.

[0071] Increase in energy consumption for attitude adjustment Related to the path turning angle and the number of adjustments, it can be expressed as: in, Energy consumption coefficient for attitude adjustment For the first The angle of the path turning point in this adjustment. To adjust the number of times.

[0072] Then minimize the extra energy consumption Represented as: min Simultaneously, constraints on time, path, and UUV performance are set: 1) Time constraint: The replanned path must be completed within a specified time. (In this embodiment, the mapping task is to be completed within 600 seconds, i.e.:) 2) Path constraints: The replanned path in the survey area model Inside, and avoiding areas with obstacles. ;Right now and .

[0073] 3) UUV Performance Constraints: The UUV's travel speed, depth, and attitude adjustment angle must meet the corresponding upper and lower limits. UUV travel speed... Must meet ,in, , These represent the minimum and maximum speeds of a UUV, and the depth. Must meet ,in, , These represent the safe operating depth range and attitude adjustment angle for UUVs. Must meet ,in, , These are the minimum and maximum attitude adjustment angles for the UUV, respectively.

[0074] S44, an improved pathfinding algorithm In this embodiment, the improved path search algorithm employs the improved Fast Random Tree Search (RRT*) algorithm.

[0075] Traditional RRT* algorithms suffer from low search efficiency due to the lack of guidance from randomly sampled points during the search process. This invention introduces a heuristic function to guide the sampling points. Based on the current node To the target point Euclidean distance and water flow influencing factors Construct, represented as: in, The value is dynamically adjusted based on the water flow velocity; the higher the flow velocity, the larger the value, guiding the algorithm to prioritize searching in directions with less influence from the water flow. In this embodiment, Based on the magnitude of the water flow velocity The adjustments are made dynamically between them.

[0076] Furthermore, during the RRT* algorithm's tree expansion process, when generating new nodes... At that time, it is not only considered in relation to its parent node. The connection cost is considered in addition to the minimum cost from the current node to an existing node. By reselecting a parent node, the new node is connected to the parent node with the lowest cost, thereby reducing the overall path cost. The specific formula is: in For the set of existing nodes, For nodes The cost to reach the target point For nodes The cost of reaching a new node. Using the improved RRT* algorithm described above, the additional energy consumption can be quickly searched while satisfying the constraints. The minimum replanning path enables local replanning of surveying operations based on an energy-saving optimization model.

[0077] Furthermore, this embodiment also provides a near-shore mapping system, implemented based on the methods described above. It integrates the aforementioned key technologies and, addressing the complex near-shore seabed environment and UUV energy constraints, achieves intelligent management of the entire process from data processing and path planning to operation monitoring, providing an efficient and energy-saving solution for UUV mapping operations. The system adopts a layered architecture design, divided into a data layer, an algorithm layer, a functional layer, and a user interaction layer. The data layer is responsible for storing and managing prior underwater 3D environmental data, satellite data, and real-time UUV data transmission, with data stored in a database in both structured and unstructured formats. The algorithm layer integrates deep learning-based algorithms for land-sea boundary identification and mapping area modeling, energy-constrained coverage planning algorithms, and energy-saving optimization-based local replanning algorithms. The functional layer implements core functions such as data processing, path planning, and operation monitoring. The user interaction layer provides an intuitive interface for convenient parameter setting, task management, and status viewing.

[0078] Specifically, such as Figure 2 As shown, the system includes: (1) A data processing module, used to perform the data acquisition and preprocessing steps and the deep learning-based land-sea boundary contour recognition and mapping area modeling steps. Details have been described above and will not be repeated here.

[0079] Furthermore, in one embodiment, the module also performs feature fusion processing on the prior data and synchronizes the real-time UUV data transmission. Specifically, For the prior data, a multi-scale feature fusion algorithm is employed. Let the satellite image feature vector be... The underwater terrain feature vector is Through weighted fusion formula The fused feature vectors provide more comprehensive information for subsequent modeling and path planning. , , Adjustments will be made dynamically based on the results of data reliability assessments.

[0080] For real-time data transmission from UUVs, including sensor data and position / attitude data, a time synchronization algorithm is employed to align different types of sensor data along the time dimension. Let the sensor data timestamp be... The timestamp for the position and attitude data is Linear interpolation algorithms are used to unify the data into the same time series for comprehensive analysis and processing. in , These are the data values ​​corresponding to adjacent timestamps. , Adjacent timestamps For the target timestamp.

[0081] (2) The path planning module is used to execute the UUV mapping operation coverage planning step based on energy constraints and the mapping operation local replanning step based on the energy-saving optimization model. The details have been described above and will not be repeated here.

[0082] (3) Operation monitoring module, used to display the UUV's location, navigation path, energy consumption, and mission progress in real time. It displays the UUV's operational trajectory within the surveyed area in a 3D map format via a visual interface, and uses different colors and icons to represent the UUV's status (e.g., normal operation, encountering obstacles, low energy). In addition, this module provides an energy consumption prediction function: based on the UUV's real-time energy consumption data and remaining path, a gray prediction model is used. Perform energy consumption prediction.

[0083] Let the recorded energy consumption sequence of UUV be... A sequence is generated by accumulation. ,in Construct differential equations ,in , For the parameters to be determined, the least squares method is used for estimation, and the prediction model is obtained by solving the problem. This allows us to obtain the predicted energy consumption value.

[0084] Based on the energy consumption prediction results, when the remaining energy of the UUV is lower than the preset threshold, the system will automatically issue an early warning and adjust the path or terminate the task according to the preset strategy.

[0085] (4) Intelligent decision-making and optimization module, used to optimize subsequent surveying and mapping tasks based on operation monitoring data and historical operation data.

[0086] In this embodiment, a reinforcement learning algorithm is employed, using task completion time, energy consumption, and mapping accuracy as reward functions to allow the software to learn the optimal decision-making strategy through continuous task execution. The reward function is set as follows: in, This is the standard task completion time. This refers to the actual time it took to complete the task. For the mission's budgeted energy consumption, For UUV energy consumption, This is a value used to assess the accuracy of the surveying and mapping. , , These are the weighting coefficients.

[0087] In this embodiment, the state space includes multi-dimensional information such as the current state of the UUV, environmental information, and mission progress, while the action space contains the relevant control quantities for UUV navigation control.

[0088] In one embodiment, the PPO (Proximity Policy Optimization) algorithm can be used to balance exploration and exploitation.

[0089] The specific implementation methods related to reinforcement learning are common practices for those skilled in the art, and will not be elaborated upon in this embodiment.

[0090] By continuously adjusting the path planning parameters and UUV operation parameters, the reward function value is maximized. This enables continuous optimization of surveying and mapping operations.

[0091] Through the coordinated operation of the above functional modules, the system achieves efficient, intelligent, and energy-saving management of UUV nearshore seabed mapping operations, and enhances the UUV's mapping capabilities in complex nearshore environments.

[0092] In the implementation of the above system, pre-processed prior underwater 3D environmental data and satellite data are first imported, and the data processing module is automatically invoked for data processing. Then, the path planning module is invoked for path planning, starting with initial path planning: an energy-constrained coverage planning algorithm is used to generate an initial mapping operation path based on the mapping area model and user-defined task parameters (such as weight coefficients and UUV performance parameters), and the path information is sent to the UUV via underwater acoustic communication. During this process, the operation monitoring module receives sensor data and position / attitude data uploaded by the UUV in real time and displays the UUV's operation status on a 3D visualization interface. When a local replanning condition is triggered, a local replanning based on energy-saving optimization is performed, generating a new path and sending it to the UUV. Simultaneously, a grey prediction model is used to predict UUV energy consumption in real time, issuing an early warning when the remaining energy is below 20%. After completing the mapping task, the UUV transmits the mapping data back to the software for processing and analysis, generating a mapping report. Meanwhile, the intelligent decision-making and optimization module adjusts parameters based on the task data using reinforcement learning algorithms, optimizing the path planning and operation strategies for subsequent tasks.

[0093] Through the above specific implementation methods, the energy efficiency of UUV coverage paths for nearshore seabed mapping can be effectively optimized, improving the efficiency and energy-saving effect of UUVs in nearshore seabed mapping operations and meeting practical application needs.

[0094] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for optimizing the energy efficiency of UUV coverage paths in near-coastal mapping, characterized in that, Includes the following steps: Data acquisition and preprocessing: acquire satellite remote sensing images, multibeam bathymetry data and underwater topographic data of nearshore areas; perform radiometric and geometric correction on satellite remote sensing images, and perform noise removal and data interpolation on multibeam bathymetry data; Deep learning-based land-sea boundary contour recognition and mapping area modeling: Preprocessed satellite remote sensing images are input into a trained deep convolutional neural network to identify land-sea boundary contours; combined with land-sea boundary contours and preprocessed multibeam bathymetry data and underwater topographic data, a three-dimensional mapping area model of the nearshore seabed is constructed. UUV mapping operation coverage planning under energy constraints: Construct a UUV energy consumption model, comprehensively consider the impact of navigation, attitude adjustment and sensor operation on energy consumption, with the goal of maximizing the coverage of the mapping area and minimizing the total energy consumption, combine the mapping area model and use intelligent optimization algorithm to perform global path planning to determine the UUV mapping operation path. Local replanning of surveying operations based on energy-saving optimization model: Real-time monitoring of the underwater environment through sensors on UUV, acquisition of obstacle information and water flow velocity data and preprocessing. Based on obstacle information and water flow velocity, the feasibility of the original planned path is determined. When the original planned path is not feasible, an improved path search algorithm is adopted with the goal of minimizing additional energy consumption, and local path replanning is performed under the constraints of time, path and UUV performance.

2. The method for optimizing UUV coverage paths for near-coastal mapping according to claim 1, characterized in that, The deep convolutional neural network adopts the U-Net network structure, which includes an encoding layer and a decoding layer. The encoding layer extracts image features from the input satellite remote sensing image through convolution and max pooling operations. The decoding layer restores the image resolution and performs feature stitching through deconvolution operations, outputting a probability map of the land-sea boundary, and obtains the land-sea boundary contour through threshold segmentation.

3. The method for optimizing UUV coverage paths for near-coastal mapping according to claim 1, characterized in that, The UUV energy consumption model is expressed as follows: in, The total energy consumption of UUV For UUV in the Energy consumption for navigation along a segment of the route. n The number of segments for the discretized path. Energy consumption for UUV attitude adjustment For sensor operating power consumption, The constant energy consumption per unit time for sensor operation. Total operation time; Assuming UUV is in the first If a segment of the path is traveled at a constant speed, then the energy consumption of that segment of the journey is... for: in, For the density of water, The underwater drag coefficient, For the cross-sectional area of ​​the UUV, For UUV in the A constant sailing speed along the route. For UUV in the The sailing distance of a segment of the route; The energy consumption for UUV attitude adjustment is expressed as: in, Energy consumption coefficient for attitude adjustment Adjust the angle to maintain the posture. To adjust the frequency.

4. The method for optimizing UUV coverage paths for near-coastal mapping according to claim 1, characterized in that, The specific method of using intelligent optimization algorithms for global path planning is as follows: based on the coverage rate of the surveyed area... Maximize and total energy consumption Minimize to a bi-objective function: The bi-objective function is transformed into a single-objective function: in, For the theoretical maximum coverage, The maximum energy that a fully charged UUV battery can provide. , These are the weighting coefficients; Using the single objective function as the optimization objective, a non-dominated sorting genetic algorithm is employed to solve the problem, obtaining the optimal path scheme under the conditions of satisfying energy consumption constraints, path constraints, speed constraints, and depth constraints.

5. The method for optimizing UUV coverage paths for near-coastal mapping according to claim 1, characterized in that, The process involves using sensors mounted on the UUV to monitor the underwater environment in real time, acquiring obstacle information and water flow velocity data, and then preprocessing the data. Obstacle information is acquired using a multi-beam sonar sensor mounted on the UUV, and the detection data from the multi-beam sonar sensor is represented as a point cloud set. ,in For points in a point cloud, These are the three-dimensional coordinates of the point. The point cloud count; water flow velocity is obtained using an acoustic Doppler current profiler mounted on the UUV, at time... The measured water flow velocity vector is ,in, These represent the water flow at... , , The velocity component in the direction; For point cloud data acquired by multibeam sonar sensors, a statistical filtering algorithm is used to remove outliers: assuming a certain point in the point cloud data... To its The average distance between nearest neighbors is ,like Greater than the global average distance With preset multiple If the product of the product of the product of the product and the product of the product, then the point is determined to be an outlier and is removed. The water flow velocity data collected by the acoustic Doppler current profiler is processed using a Kalman filter algorithm to predict the water flow velocity at the next moment.

6. The method for optimizing UUV coverage paths for near-coastal mapping according to claim 1, characterized in that, The minimized additional energy consumption is expressed as: min in, For additional energy consumption, Increase in navigation energy consumption To account for the increase in energy consumption for attitude adjustment, let the original planned path be... The replanned path is ,but in, , These represent the number of path segments before and after the replanning. , The first and second phases of the replanning were respectively The location coordinates of the path segment. For the density of water, The underwater drag coefficient, For the cross-sectional area of ​​the UUV, The velocity of the UUV relative to the ground. , For the UUV's own travel speed, For water flow velocity, For the first Travel time for a segment of the route , This is the length of the path segment. For the replanned path of UUV The magnitude of the velocity relative to the ground at any given moment. This represents the speed of the UUV relative to the ground at the corresponding location and time along its original planned path. Energy consumption coefficient for attitude adjustment For the first The angle of the path turning point in this adjustment. To adjust the number of times.

7. The method for optimizing UUV coverage paths for near-coastal mapping according to claim 1, characterized in that, The constraints on time, path, and UUV performance are expressed as follows: Time constraint: The replanned path must be completed within a specified time. The surveying task was completed within the specified timeframe, namely: in, For the replanned path of UUV The magnitude of the velocity relative to the ground at any given moment. For the first Length of the path segment The number of path segments after replanning; Path constraints: The replanned path must be within the survey area model and avoid obstacle areas; UUV performance constraints: The speed, depth, and attitude adjustment angle of the UUV must meet the corresponding upper and lower limits.

8. The method for optimizing UUV coverage paths for near-coastal mapping according to claim 1, characterized in that, The improved path search algorithm employs an improved fast search random tree algorithm, introducing a heuristic function to guide the search of sampling points. During tree expansion, it reselects parent nodes to connect new nodes to the parent node with the lowest cost, thereby reducing the overall path cost. The heuristic function... Based on the current node To the target point Euclidean distance and water flow influencing factors Construct, represented as: in, The value is dynamically adjusted based on the water flow velocity; the greater the water flow velocity, the larger the value.

9. A near-coastal mapping system, characterized in that, Based on the method described in any one of claims 1-8, the system comprises: The data processing module is used to perform the data acquisition and preprocessing steps and the deep learning-based land-sea boundary contour recognition and mapping area modeling steps. The path planning module is used to execute the UUV mapping operation coverage planning step based on energy constraints and the mapping operation local replanning step based on the energy-saving optimization model. The operation monitoring module is used to display the UUV's location, navigation path, energy consumption, and mission progress information in real time. It displays the UUV's operation trajectory in the survey area in the form of a 3D map through a visual interface. Based on the UUV's real-time energy consumption data and remaining path, it uses a gray prediction model to predict energy consumption. According to the energy consumption prediction results, when the UUV's remaining energy is lower than a preset threshold, the system automatically issues an early warning and adjusts the path or terminates the mission according to preset strategies.

10. A near-shore mapping system according to claim 9, characterized in that, The system also includes an intelligent decision-making and optimization module, which employs a reinforcement learning algorithm and sets the reward function as follows: in, This is the standard task completion time. This refers to the actual time it took to complete the task. For the mission's budgeted energy consumption, For UUV energy consumption, This is a value used to assess the accuracy of the surveying and mapping. , , These are the weighting coefficients; By continuously adjusting the path planning parameters and UUV operation parameters, the reward function value is maximized. This enables continuous optimization of surveying and mapping operations.

Images