Underwater robot hydrodynamic simulation method and system

Abstract

The application relates to the technical field of water power simulation, and particularly discloses a water power simulation method and system for an underwater robot, which comprises the following steps: setting pressure detection points on the surface of a robot model; acquiring real-time pressure containing positions and time based on the pressure detection points, and constructing pressure samples; clustering the pressure samples, querying pose parameters and power parameters of the underwater robot according to the time of the pressure samples for each type of pressure sample, and constructing a regression model of the power parameters to the pose parameters; for the real-time acquired pressure samples, determining target type pressure samples in the clustering results, reading the regression model corresponding to the target type pressure samples, and performing water power simulation on the power parameters; the application determines environmental conditions based on sensing points, reads the regression model to perform water power simulation, and the workload of the control module is only reading data, applying functions and outputting instructions; under the premise of a certain accuracy, the response speed is extremely fast.

Description

Technical Field

[0001] This invention relates to the field of hydrodynamic simulation technology, specifically to a method and system for simulating the hydrodynamics of an underwater robot. Background Technology

[0002] Hydrodynamic simulation refers to the process of modeling and simulating the hydrodynamic behavior of underwater vehicles (such as underwater robots, submarines, and surface ships) in water using computational methods. The purpose is to predict key indicators such as forces, motion response, propulsion efficiency, attitude changes, and energy consumption, providing a basis for design and control.

[0003] Existing technologies include high-performance finite element analysis software that can be used for hydrodynamic simulation. However, this requires a large amount of computing resources and is obviously not feasible to install on the control module of an underwater robot, due to both memory limitations and insufficient computing speed. Therefore, the technical problem that this invention aims to solve is to provide a lightweight hydrodynamic simulation solution for underwater robots. Summary of the Invention

[0004] The purpose of this invention is to provide a hydrodynamic simulation method and system for underwater robots to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method and system for simulating the hydrodynamics of an underwater robot, the method comprising:

[0007] Obtain a robot model of the underwater robot and set pressure detection points on the surface of the robot model;

[0008] Pressure samples are constructed by acquiring instantaneous pressure containing location and time information in real time based on pressure detection points;

[0009] The pressure samples are clustered, and for each pressure sample, the pose and dynamic parameters of the underwater robot are queried according to the time of the pressure sample. A regression model from the dynamic parameters to the pose parameters is constructed. The number of indicators for the pose parameters and the dynamic parameters are preset values.

[0010] For the pressure samples acquired in real time, the target class pressure samples are determined from the clustering results, the regression model corresponding to the target class pressure samples is read, and hydrodynamic simulation is performed on the dynamic parameters.

[0011] As a further aspect of the present invention: the step of obtaining a robot model of an underwater robot and setting pressure detection points on the surface of the robot model includes:

[0012] Obtain a robot model of the underwater robot;

[0013] Insert a cover mesh of preset precision into the surface of the robot model and obtain the feature values ​​of each position in the cover mesh;

[0014] The locations where the feature values ​​reach the preset feature value threshold are selected as pressure detection points;

[0015] The independent variables in the feature value calculation process include two types: one is the distance between the position and the robot model, and the other is the distance between the position and the selected position.

[0016] As a further aspect of the present invention: the step of constructing a pressure sample by real-time acquisition of instantaneous pressure containing location and time based on pressure detection points includes:

[0017] Instantaneous pressure is acquired using detection equipment installed at pressure detection points, and the location of the pressure detection points and the time of acquisition of the instantaneous pressure are recorded.

[0018] The instantaneous pressure at all locations at the same acquisition time is statistically analyzed to construct an instantaneous pressure table; the instantaneous pressure table includes a location item and an instantaneous pressure item;

[0019] For any instantaneous pressure gauge, compare the instantaneous pressure gauge with the previous instantaneous pressure gauge and the next instantaneous pressure gauge respectively, and determine the first-order difference and second-order difference of the data at each position.

[0020] The weight of each position is determined based on the first-order difference and the second-order difference.

[0021] Construct weight terms, insert instantaneous pressure gauges, and obtain pressure samples.

[0022] As a further aspect of the present invention: the step of clustering the pressure samples, and for each class of pressure samples, querying the underwater robot's pose and dynamic parameters based on the time of the pressure sample, and constructing a regression model from dynamic parameters to pose parameters includes:

[0023] The obtained pressure samples are compared pairwise to calculate the sample distance between any two pressure samples.

[0024] The pressure samples are clustered based on the distance to obtain different classes of pressure samples.

[0025] For each type of pressure sample, the time of each pressure sample is queried in turn, and the pose parameters and dynamic parameters at that time are queried in the log of the underwater robot as parameter samples.

[0026] For any index in the pose parameters, construct a regression model for that index based on the dynamic parameters;

[0027] The regression model of all indicators in the statistical pose parameters is used as the regression model from the dynamic parameters to the pose parameters of this type of pressure sample.

[0028] As a further aspect of the present invention: the step of comparing the obtained pressure samples pairwise to calculate the sample distance between any two pressure samples includes:

[0029] The obtained pressure samples are paired up in pairs;

[0030] For two paired stress samples, positions are selected in descending order of weight, and the data at those positions is compared to calculate the data distance.

[0031] The accumulated data distances are used to set the sample distances to a preset maximum value when the data distances reach a preset threshold.

[0032] Once the data for all locations has been calculated, the cumulative data distance is used as the sample distance.

[0033] As a further aspect of the present invention: the steps of determining the target class pressure sample from the clustering results, reading the regression model corresponding to the target class pressure sample, and performing hydrodynamic simulation on the dynamic parameters for the pressure sample acquired in real time include:

[0034] Read the latest acquired pressure sample, compare it with various types of pressure samples, and calculate the matching degree;

[0035] Select a class of stress samples whose matching degree reaches a preset matching degree threshold as the target class of stress samples;

[0036] Read the regression model of the target class pressure sample, and perform hydrodynamic simulation on the current dynamic parameters based on the regression model;

[0037] The accuracy of hydrodynamic simulation results is calculated by periodically verifying them using AI, and the application frequency of AI is adjusted based on the accuracy. The application frequency is inversely proportional to the accuracy.

[0038] The present invention also provides an underwater robot hydrodynamic simulation system, the system comprising:

[0039] The pressure point creation module is used to acquire the robot model of the underwater robot and set pressure detection points on the surface of the robot model.

[0040] The pressure sample construction module is used to acquire instantaneous pressure containing location and time information in real time based on pressure detection points and construct pressure samples.

[0041] The regression model creation module is used to cluster the pressure samples. For each pressure sample, the underwater robot's pose and dynamic parameters are queried based on the time of the pressure sample, and a regression model from the dynamic parameters to the pose parameters is constructed. The number of indicators for the pose parameters and the dynamic parameters are preset values.

[0042] The sample simulation module is used to determine the target class pressure sample from the clustering results of the pressure sample acquired in real time, read the regression model corresponding to the target class pressure sample, and perform hydrodynamic simulation on the dynamic parameters.

[0043] As a further aspect of the present invention: the pressure point creation module includes:

[0044] The model acquisition unit is used to acquire the robot model of the underwater robot.

[0045] The eigenvalue calculation unit is used to insert a cover mesh of preset precision into the surface of the robot model and obtain the eigenvalues ​​at each position in the cover mesh;

[0046] The location-first selection unit is used to select the location where the feature value reaches the preset feature value threshold as the pressure detection point;

[0047] The independent variables in the feature value calculation process include two types: one is the distance between the position and the robot model, and the other is the distance between the position and the selected position.

[0048] As a further aspect of the present invention: the pressure sample construction module includes:

[0049] The data acquisition unit is used to acquire instantaneous pressure based on the detection equipment installed at the pressure detection point, and to record the location of the pressure detection point and the acquisition time of the instantaneous pressure.

[0050] A pressure gauge construction unit is used to statistically analyze the instantaneous pressure at all locations at the same acquisition time and construct an instantaneous pressure gauge; the instantaneous pressure gauge includes a location item and an instantaneous pressure item;

[0051] The differential calculation unit is used to compare any instantaneous pressure gauge with the previous and next instantaneous pressure gauges respectively, and determine the first-order and second-order differences of the data at each position.

[0052] The weight determination unit is used to determine the weight of each position based on the first-order difference and the second-order difference;

[0053] The weight term construction unit is used to construct weight terms, insert instantaneous pressure gauges, and obtain pressure samples.

[0054] As a further aspect of the present invention: the regression model creation module includes:

[0055] The sample distance calculation unit is used to compare the acquired pressure samples pairwise and calculate the sample distance between any two pressure samples.

[0056] A sample clustering unit is used to cluster pressure samples based on the distance to obtain pressure samples of different classes.

[0057] The parameter sample generation unit is used to query the time of each pressure sample in turn for each type of pressure sample, and query the pose parameters and dynamic parameters at that time in the underwater robot's log as parameter samples.

[0058] Construct an execution unit to build a regression model for any index in the pose parameters based on the dynamic parameters;

[0059] The statistical unit is used to statistically analyze the regression model of all indicators in the pose parameters, serving as a regression model from the dynamic parameters to the pose parameters for this type of pressure sample.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] This invention installs multiple sensor points on an underwater robot to acquire different environmental conditions. For the same environmental condition, it reads the pose and dynamic parameters under that condition and constructs a mapping relationship between the dynamic parameters and the pose parameters. At this point, the mapping relationship itself is a function. In actual use, the environmental conditions are determined based on the sensor points, and the mapping relationship is read to perform hydrodynamic simulation. The workload of the underwater robot's control module is only to read data, apply functions, and output commands. With a certain degree of accuracy, the response speed is extremely fast. Attached Figure Description

[0062] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0063] Figure 1 The overall flowchart of the underwater robot hydrodynamic simulation method is shown.

[0064] Figure 2 The structural diagram of the underwater robot hydrodynamic simulation system is shown. Detailed Implementation

[0065] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0066] Figure 1 This is a flowchart illustrating the overall process of an underwater robot hydrodynamic simulation method and system. In this embodiment of the invention, an underwater robot hydrodynamic simulation method includes:

[0067] Step S100: Obtain the robot model of the underwater robot and set pressure detection points on the surface of the robot model;

[0068] When designing an underwater robot, there is a component model called a robot model. In the technical solution of this invention, the robot model is considered to be known data. The robot model is analyzed, pressure detection points are set on the surface of the robot model, and patch sensors are installed at the pressure detection points to obtain underwater pressure or force.

[0069] Step S200: Based on the pressure detection points, acquire the instantaneous pressure containing location and time in real time, and construct a pressure sample;

[0070] A patch pressure sensor is installed at the pressure detection point to acquire underwater pressure. When acquiring underwater pressure, the instantaneous pressure is obtained, and the location and time need to be saved. A single acquisition operation can obtain the instantaneous pressure at all locations at that time, which is called a pressure sample.

[0071] Step S300: Cluster the pressure samples. For each pressure sample, query the underwater robot's pose parameters and dynamic parameters based on the time of the pressure sample, and construct a regression model from dynamic parameters to pose parameters. The number of indicators for the pose parameters and dynamic parameters are preset values.

[0072] The pressure sample acquisition process is real-time. When the underwater robot is placed in the underwater environment, the sensors installed at the pressure detection points work in real time, resulting in a large number of pressure samples. In addition, multiple underwater robots can be used during the test and simulation phase, thus obtaining even more pressure samples. All the obtained pressure samples are clustered. After clustering, each type of pressure sample is analyzed separately. The pose and dynamic parameters of the underwater robot are queried based on the time of each pressure sample. The pose and dynamic parameters are obtained from relevant existing modules, which are the basic modules of the underwater robot. The obtained data is also considered as default known data. Then, a mapping relationship from dynamic parameters to pose parameters is constructed. The mapping relationship constructed by the technical solution of this invention is a regression model.

[0073] Specifically, the number of indicators for both the pose parameters and the power parameters are preset values. The pose parameters mainly include attitude parameters, which describe the robot's orientation or rotation state. They are commonly represented by Euler angles or quaternions, including: roll (rotation around the x-axis, left and right roll); pitch (rotation around the y-axis, up and down nodding); and yaw (rotation around the z-axis, horizontal turning). The power parameters are the source power of each power component. The types and number of power components of an underwater robot are not unique, but they are not too many, and they are used to adjust the attitude parameters.

[0074] Step S400: For the pressure samples acquired in real time, determine the target class pressure samples in the clustering results, read the regression model corresponding to the target class pressure samples, and perform hydrodynamic simulation on the dynamic parameters.

[0075] Step S400 is the application process. For the pressure sample acquired in real time, it is compared with each type of pressure sample to determine which category the real-time pressure sample belongs to, called the target pressure sample. The regression model corresponding to the target pressure sample is read, and the dynamic parameters to be used are obtained. The dynamic parameters are input into the regression model to obtain the predicted pose parameters. The function of this architecture is that when pose adjustment is required, some dynamic parameters can be generated first. With the help of the obtained regression model, the pose corresponding to each dynamic parameter can be predicted to assist the self-adjustment process of the underwater robot, introducing a simulation function with very low resource requirements to the underwater robot.

[0076] Regarding step S100, the step of obtaining the robot model of the underwater robot and setting pressure detection points on the surface of the robot model includes:

[0077] Obtain a robot model of the underwater robot;

[0078] Insert a cover mesh of preset precision into the surface of the robot model and obtain the feature values ​​of each position in the cover mesh;

[0079] The locations where the feature values ​​reach the preset feature value threshold are selected as pressure detection points.

[0080] The underwater robot model is a known model and can be directly obtained. A cover mesh with a preset precision is inserted into the surface of the robot model. The cover mesh is simply a mesh inserted into the surface of the robot model. It can be that a mesh is generated first and then covered onto the surface of the robot model, hence the name cover mesh. Mesh generation is a basic function in many existing software programs, so it will not be elaborated here. The precision corresponds to the cell size of the mesh. The smaller the cell size, the higher the precision. Based on this, the feature values ​​of each position in the cover mesh are obtained, and the positions where the feature values ​​reach the preset feature value threshold are selected as pressure detection points.

[0081] The independent variables in the eigenvalue calculation process include two types: the distance between the position and the robot model, and the distance between the position and the selected position. One feasible approach is as follows:

[0082] The robot model is segmented using a 3D mesh of a different precision to obtain multiple 3D points within the robot model. The importance of the component at each 3D point is queried (to characterize the importance of the component). For any position on the covering mesh, its distance to all 3D points is calculated. The importance of each 3D point is divided by its distance, and then the results are summed to obtain the importance of each position in the covering mesh. The closer a position is to an important component, the higher its importance. Based on this, the nearest selected position is queried, and its distance is calculated. The importance is divided by this distance (an adjustment coefficient can be introduced). The resulting value is called the eigenvalue. In summary, the closer a position is to an important component, the farther away the nearest selected position is, the larger the eigenvalue of that position, and the more likely it is to be selected. For a certain important component, the number of selected points is also greater.

[0083] Regarding step S200, the step of constructing a pressure sample by acquiring instantaneous pressure containing location and time in real time based on pressure detection points includes:

[0084] Instantaneous pressure is acquired using detection equipment installed at pressure detection points, and the location of the pressure detection points and the time of acquisition of the instantaneous pressure are recorded.

[0085] The instantaneous pressure at all locations at the same acquisition time is statistically analyzed to construct an instantaneous pressure table; the instantaneous pressure table includes a location item and an instantaneous pressure item;

[0086] For any instantaneous pressure gauge, compare the instantaneous pressure gauge with the previous instantaneous pressure gauge and the next instantaneous pressure gauge respectively, and determine the first-order difference and second-order difference of the data at each position.

[0087] The weight of each position is determined based on the first-order difference and the second-order difference.

[0088] Construct weight terms, insert instantaneous pressure gauges, and obtain pressure samples.

[0089] The above describes the construction process of the pressure sample. Instantaneous pressure is acquired using a detection device installed at the pressure detection point. This detection device is a pressure sensor. When acquiring instantaneous pressure, it is also necessary to record the time and location. Then, using time as an index, the instantaneous pressure at all locations at the same acquisition time is statistically analyzed to construct an instantaneous pressure table. When the time difference between the instantaneous pressure at two locations is sufficiently small, they are considered to belong to the same acquisition time. For any instantaneous pressure table, it is compared with the previous and subsequent instantaneous pressure tables to determine the first-order and second-order differences of the data at each location. The calculation process for the first-order difference is as follows: calculate the difference between the current data and the previous data, calculate the difference between the subsequent data and the current data, and then calculate the mean to obtain the first-order difference. The calculation process for the second-order difference is as follows: calculate the sum of the subsequent data and the previous data, and then subtract twice the current data to obtain the second-order difference.

[0090] The weight of each position is determined based on the first-order and second-order differences, a weight term is constructed, and an instantaneous pressure gauge is inserted to obtain a pressure sample; wherein, the weight term is proportional to both the first-order and second-order differences.

[0091] Regarding step S300, the step of clustering the pressure samples, and for each pressure sample, querying the underwater robot's pose and dynamic parameters based on the time of the pressure sample, and constructing a regression model from dynamic parameters to pose parameters includes:

[0092] The obtained pressure samples are compared pairwise to calculate the sample distance between any two pressure samples.

[0093] The pressure samples are clustered based on the distance to obtain different classes of pressure samples.

[0094] For each type of pressure sample, the time of each pressure sample is queried in turn, and the pose parameters and dynamic parameters at that time are queried in the log of the underwater robot as parameter samples.

[0095] For any index in the pose parameters, construct a regression model for that index based on the dynamic parameters;

[0096] The regression model of all indicators in the statistical pose parameters is used as the regression model from the dynamic parameters to the pose parameters of this type of pressure sample.

[0097] The number of pressure samples obtained is large. Each pressure sample is compared pairwise to calculate the sample distance between any two pressure samples. Based on this distance, the pressure samples are clustered to obtain different classes of pressure samples. The clustering scheme used in this invention is an unlimited-number clustering scheme, such as the OPTICS clustering algorithm. After clustering, each class of pressure samples is analyzed separately. For each class of pressure samples, the time of each pressure sample is queried sequentially. The pose parameters and dynamic parameters at that time are retrieved from the underwater robot's log. The pose parameters and dynamic parameters at that time constitute a data pair, called a parameter sample. One time corresponds to one parameter sample. The pose parameters are the dependent variable, and the dynamic parameters are the independent variables. To make the processing clearer, each indicator in the pose parameters is used as the dependent variable. A regression model from the dynamic parameters to each indicator is trained. The regression models of all indicators in the pose parameters are statistically analyzed and used as the regression model from the dynamic parameters to the pose parameters for that class of pressure samples.

[0098] Specifically, the step of comparing the acquired pressure samples pairwise to calculate the sample distance between any two pressure samples includes:

[0099] The obtained pressure samples are paired up in pairs;

[0100] For two paired stress samples, positions are selected in descending order of weight, and the data at those positions is compared to calculate the data distance.

[0101] The accumulated data distances are used to set the sample distances to a preset maximum value when the data distances reach a preset threshold.

[0102] Once the data for all locations has been calculated, the cumulative data distance is used as the sample distance.

[0103] In one example of the technical solution of this invention, the calculation process of sample distance is described. The obtained pressure samples are paired up in pairs. For the two paired pressure samples, positions are selected according to the descending order of weights. The data at the corresponding positions is compared, and the data distance is calculated. The data distance can be the product of the weight and the absolute value of the difference. The accumulated data distance is used. When the data distance is large enough, the two pressure samples can be considered as not being of the same type. The sample distance is then set to a preset maximum value. After the data at all positions has been calculated, the accumulated data distance is used as the sample distance. The advantage of introducing this process is that important data is analyzed first, and obviously different pressure samples are not analyzed further, thus improving the comparison efficiency.

[0104] Regarding step S400, the steps of determining the target class pressure sample from the clustering results, reading the regression model corresponding to the target class pressure sample, and performing hydrodynamic simulation on the dynamic parameters for the pressure sample acquired in real time include:

[0105] Read the latest acquired pressure sample, compare it with various types of pressure samples, and calculate the matching degree;

[0106] Select a class of stress samples whose matching degree reaches a preset matching degree threshold as the target class of stress samples;

[0107] Read the regression model of the target class pressure sample, and perform hydrodynamic simulation on the current dynamic parameters based on the regression model;

[0108] The accuracy of hydrodynamic simulation results is calculated by periodically verifying them using AI, and the application frequency of AI is adjusted based on the accuracy. The application frequency is inversely proportional to the accuracy.

[0109] In one example of the technical solution of the present invention, the practical application stage is described. The latest pressure sample is read and compared with various pressure samples to calculate the matching degree. A type of pressure sample with a matching degree that reaches a preset matching degree threshold is selected as the target pressure sample. The regression model of the target pressure sample is read and hydrodynamic simulation is performed on the current dynamic parameters based on the regression model.

[0110] Based on this, AI is introduced to verify the hydrodynamic simulation results and determine the accuracy of the hydrodynamic simulation results. The AI ​​verification process is slower but more accurate. The application frequency of AI is adjusted according to the accuracy. If the accuracy is too low, the hydrodynamic simulation process of this invention may have failed. In this case, the application frequency of AI should be higher. Conversely, if the accuracy is high, the application frequency of AI should be lower.

[0111] In addition, regarding the matching degree parameter, one feasible approach is to iterate through all types of pressure samples using the latest acquired pressure sample, calculate the similarity in real time, and select the maximum similarity as the matching degree. Since the pressure samples adopt a table structure, the similarity can be calculated by applying the similarity calculation process of the table structure.

[0112] Figure 2 A structural diagram of an underwater robot hydrodynamic simulation system is shown. In a preferred embodiment of the technical solution of the present invention, an underwater robot hydrodynamic simulation system is also provided, the system 10 comprising:

[0113] Pressure point creation module 11 is used to acquire the robot model of the underwater robot and set pressure detection points on the surface of the robot model;

[0114] The pressure sample construction module 12 is used to acquire instantaneous pressure containing location and time in real time based on pressure detection points and construct pressure samples.

[0115] The regression model creation module 13 is used to cluster the pressure samples. For each type of pressure sample, the underwater robot's pose parameters and dynamic parameters are queried according to the time of the pressure sample, and a regression model from dynamic parameters to pose parameters is constructed. The number of indicators for the pose parameters and the dynamic parameters are preset values.

[0116] The sample simulation module 14 is used to determine the target class pressure sample in the clustering results for the pressure sample acquired in real time, read the regression model corresponding to the target class pressure sample, and perform hydrodynamic simulation on the dynamic parameters.

[0117] Furthermore, the pressure point creation module 11 includes:

[0118] The model acquisition unit is used to acquire the robot model of the underwater robot.

[0119] The eigenvalue calculation unit is used to insert a cover mesh of preset precision into the surface of the robot model and obtain the eigenvalues ​​at each position in the cover mesh;

[0120] The location-first selection unit is used to select the location where the feature value reaches the preset feature value threshold as the pressure detection point;

[0121] The independent variables in the feature value calculation process include two types: one is the distance between the position and the robot model, and the other is the distance between the position and the selected position.

[0122] Specifically, the pressure sample construction module 12 includes:

[0123] The data acquisition unit is used to acquire instantaneous pressure based on the detection equipment installed at the pressure detection point, and to record the location of the pressure detection point and the acquisition time of the instantaneous pressure.

[0124] A pressure gauge construction unit is used to statistically analyze the instantaneous pressure at all locations at the same acquisition time and construct an instantaneous pressure gauge; the instantaneous pressure gauge includes a location item and an instantaneous pressure item;

[0125] The differential calculation unit is used to compare any instantaneous pressure gauge with the previous and next instantaneous pressure gauges respectively, and determine the first-order and second-order differences of the data at each position.

[0126] The weight determination unit is used to determine the weight of each position based on the first-order difference and the second-order difference;

[0127] The weight term construction unit is used to construct weight terms, insert instantaneous pressure gauges, and obtain pressure samples.

[0128] Furthermore, the regression model creation module 13 includes:

[0129] The sample distance calculation unit is used to compare the acquired pressure samples pairwise and calculate the sample distance between any two pressure samples.

[0130] A sample clustering unit is used to cluster pressure samples based on the distance to obtain pressure samples of different classes.

[0131] The parameter sample generation unit is used to query the time of each pressure sample in turn for each type of pressure sample, and query the pose parameters and dynamic parameters at that time in the underwater robot's log as parameter samples.

[0132] Construct an execution unit to build a regression model for any index in the pose parameters based on the dynamic parameters;

[0133] The statistical unit is used to statistically analyze the regression model of all indicators in the pose parameters, serving as a regression model from the dynamic parameters to the pose parameters for this type of pressure sample.

[0134] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for simulating the hydrodynamics of an underwater robot, characterized in that, The method includes: Obtain a robot model of the underwater robot and set pressure detection points on the surface of the robot model; Pressure samples are constructed by acquiring instantaneous pressure containing location and time information in real time based on pressure detection points; The pressure samples are clustered, and for each pressure sample, the pose and dynamic parameters of the underwater robot are queried according to the time of the pressure sample. A regression model from the dynamic parameters to the pose parameters is constructed. The number of indicators for the pose parameters and the dynamic parameters are preset values. For the pressure samples acquired in real time, the target class pressure samples are determined from the clustering results, the regression model corresponding to the target class pressure samples is read, and hydrodynamic simulation is performed on the dynamic parameters. The step of constructing a pressure sample by acquiring instantaneous pressure containing location and time information in real time based on pressure detection points includes: Instantaneous pressure is acquired using detection equipment installed at pressure detection points, and the location of the pressure detection points and the time of acquisition of the instantaneous pressure are recorded. The instantaneous pressure at all locations at the same acquisition time is statistically analyzed to construct an instantaneous pressure table; the instantaneous pressure table includes a location item and an instantaneous pressure item; For any instantaneous pressure gauge, compare the instantaneous pressure gauge with the previous instantaneous pressure gauge and the next instantaneous pressure gauge respectively, and determine the first-order difference and second-order difference of the data at each position. The weight of each position is determined based on the first-order difference and the second-order difference. Construct weight terms, insert instantaneous pressure gauges, and obtain pressure samples; The steps of clustering the pressure samples, and for each pressure sample, querying the underwater robot's pose and dynamic parameters based on the time of the pressure sample, and constructing a regression model from dynamic parameters to pose parameters include: The obtained pressure samples are compared pairwise to calculate the sample distance between any two pressure samples. The pressure samples are clustered based on the distance to obtain different classes of pressure samples. For each type of pressure sample, the time of each pressure sample is queried in turn, and the pose parameters and dynamic parameters at that time are queried in the log of the underwater robot as parameter samples. For any index in the pose parameters, construct a regression model for that index based on the dynamic parameters; The regression model of all indicators in the statistical pose parameters is used as the regression model from the dynamic parameters to the pose parameters of this type of pressure sample.

2. The underwater robot hydrodynamic simulation method according to claim 1, characterized in that, The step of obtaining a robot model of the underwater robot and setting pressure detection points on the surface of the robot model includes: Obtain a robot model of the underwater robot; Insert a cover mesh of preset precision into the surface of the robot model and obtain the feature values ​​of each position in the cover mesh; The locations where the feature values ​​reach the preset feature value threshold are selected as pressure detection points; The independent variables in the feature value calculation process include two types: one is the distance between the position and the robot model, and the other is the distance between the position and the selected position.

3. The underwater robot hydrodynamic simulation method according to claim 1, characterized in that, The step of comparing the obtained pressure samples pairwise to calculate the sample distance between any two pressure samples includes: The obtained pressure samples are paired up in pairs; For two paired stress samples, positions are selected in descending order of weight, and the data at those positions is compared to calculate the data distance. The accumulated data distances are used to set the sample distances to a preset maximum value when the data distances reach a preset threshold. Once the data for all locations has been calculated, the cumulative data distance is used as the sample distance.

4. The underwater robot hydrodynamic simulation method according to claim 1, characterized in that, The steps of determining the target class pressure sample from the clustering results, reading the regression model corresponding to the target class pressure sample, and performing hydrodynamic simulation on the dynamic parameters for the pressure sample acquired in real time include: Read the latest acquired pressure sample, compare it with various types of pressure samples, and calculate the matching degree; Select a class of stress samples whose matching degree reaches a preset matching degree threshold as the target class of stress samples; Read the regression model of the target class pressure sample, and perform hydrodynamic simulation on the current dynamic parameters based on the regression model; The accuracy of hydrodynamic simulation results is calculated by periodically verifying them using AI, and the application frequency of AI is adjusted based on the accuracy. The application frequency is inversely proportional to the accuracy.

5. A hydrodynamic simulation system for an underwater robot, characterized in that, The system includes: The pressure point creation module is used to acquire the robot model of the underwater robot and set pressure detection points on the surface of the robot model. The pressure sample construction module is used to acquire instantaneous pressure containing location and time information in real time based on pressure detection points and construct pressure samples. The regression model creation module is used to cluster the pressure samples. For each pressure sample, the underwater robot's pose and dynamic parameters are queried based on the time of the pressure sample, and a regression model from the dynamic parameters to the pose parameters is constructed. The number of indicators for the pose parameters and the dynamic parameters are preset values. The sample simulation module is used to determine the target class pressure sample from the clustering results of the pressure sample acquired in real time, read the regression model corresponding to the target class pressure sample, and perform hydrodynamic simulation on the dynamic parameters. The pressure sample construction module includes: The data acquisition unit is used to acquire instantaneous pressure based on the detection equipment installed at the pressure detection point, and to record the location of the pressure detection point and the acquisition time of the instantaneous pressure. A pressure gauge construction unit is used to statistically analyze the instantaneous pressure at all locations at the same acquisition time and construct an instantaneous pressure gauge; the instantaneous pressure gauge includes a location item and an instantaneous pressure item; The differential calculation unit is used to compare any instantaneous pressure gauge with the previous and next instantaneous pressure gauges respectively, and determine the first-order and second-order differences of the data at each position. The weight determination unit is used to determine the weight of each position based on the first-order difference and the second-order difference; The weight term construction unit is used to construct weight terms, insert instantaneous pressure gauges, and obtain pressure samples. The regression model creation module includes: The sample distance calculation unit is used to compare the acquired pressure samples pairwise and calculate the sample distance between any two pressure samples. A sample clustering unit is used to cluster pressure samples based on the distance to obtain pressure samples of different classes. The parameter sample generation unit is used to query the time of each pressure sample in turn for each type of pressure sample, and query the pose parameters and dynamic parameters at that time in the underwater robot's log as parameter samples. Construct an execution unit to build a regression model for any index in the pose parameters based on the dynamic parameters; The statistical unit is used to statistically analyze the regression model of all indicators in the pose parameters, serving as a regression model from the dynamic parameters to the pose parameters for this type of pressure sample.

6. The underwater robot hydrodynamic simulation system according to claim 5, characterized in that, The pressure point creation module includes: The model acquisition unit is used to acquire the robot model of the underwater robot. The eigenvalue calculation unit is used to insert a cover mesh of preset precision into the surface of the robot model and obtain the eigenvalues ​​at each position in the cover mesh; The location-first selection unit is used to select the location where the feature value reaches the preset feature value threshold as the pressure detection point; The independent variables in the feature value calculation process include two types: one is the distance between the position and the robot model, and the other is the distance between the position and the selected position.

Images