Large-scale multi-joint hydraulic mechanical arm for constructing marine ranching and control method

Abstract

The invention relates to the technical field of modernized marine ranch construction equipment, in particular to a multi-purpose large-scale multi-joint hydraulic mechanical arm capable of being universally used in an application scene and a control method. A preset working track of the mechanical arm is obtained according to the marine environment and application, key track position evaluation nodes are extracted, and a preset pose three-dimensional model diagram is constructed and stored in a feature database. In the actual operation process, the system judges the position node of the mechanical arm through real-time position analysis and matching, and carries out pose evaluation. And if the pose deviation is found, the system quickly finds an optimal deviation correction scheme according to the knowledge graph of the historical deviation correction parameters and the optimal deviation correction parameters, and accurate deviation correction is carried out. By means of the method, the operation precision and efficiency of the mechanical arm can be remarkably improved, efficient and accurate deviation correction processing can be achieved through historical data and model matching, and therefore it is guaranteed that multiple types of tasks in a modern marine ranch are conducted smoothly, and the stability and reliability of a system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of modern marine ranch construction equipment, in particular to a large-scale multi-joint hydraulic mechanical arm and a control method for building a marine ranch. Background Art

[0002] The modern marine ranch industry requires a series of high-precision and high-complexity marine operations, such as installing and repairing aquaculture cages, cleaning nets, catching fish, cleaning up seabed garbage and algae, repairing damaged coral reefs, and monitoring the marine environment. These operations often need to be carried out in complex marine environments, and the durability and safety requirements of the operating equipment are extremely high. Traditional marine ranch construction work mainly relies on manual labor, such as manual offshore operations or simple operations using small remote-controlled mechanical devices. However, this method has disadvantages such as low efficiency, high cost, and high safety risks.

[0003] As marine fisheries develop further into the deep sea, equipment is becoming larger and larger, and large-scale multi-joint hydraulic manipulators will gradually become important mechanized support devices in the field of modern marine ranching. With its powerful force, flexible movements, high precision, and durable safety, hydraulic manipulators can effectively replace manual seabed operations, improve efficiency, reduce costs, and ensure safety. However, due to the complexity and uncertainty of the marine environment, as well as the complexity of the manipulator's own structure and transmission accuracy, how to control the manipulator to perform construction work accurately, stably, and efficiently has become an urgent problem to be solved. Summary of the invention

[0004] The present invention overcomes the shortcomings of the prior art and provides a large-scale multi-joint hydraulic mechanical arm and a control method for building a marine ranch.

[0005] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is:

[0006] The first aspect of the present invention discloses a control method for a large-scale multi-joint hydraulic mechanical arm for building a marine ranch, comprising the following steps:

[0007] Obtaining a preset working trajectory of the target robotic arm, performing feature extraction on the preset working trajectory, obtaining a number of trajectory position evaluation nodes of the target robotic arm, and constructing a preset posture three-dimensional model diagram of the target robotic arm at each trajectory position evaluation node;

[0008] Constructing a blank database, and storing each preset posture three-dimensional model image in the blank database to obtain a feature database;

[0009] Performing a position analysis on the real-time position of the target robotic arm in combination with the feature database to determine whether the real-time position of the target robotic arm is a trajectory position evaluation node; if it is a trajectory position evaluation node, further analyzing the posture state of the target robotic arm at the current position node;

[0010] If the posture state of the target robot arm at the current position node is normal, no correction processing is performed on the target robot arm; if the posture state of the target robot arm at the current position node is offset, correction processing is performed on the target robot arm.

[0011] Preferably, a preset working trajectory of the target manipulator is obtained, and features are extracted for the preset working trajectory to obtain several trajectory position evaluation nodes of the target manipulator, and a preset posture three-dimensional model diagram of the target manipulator at each trajectory position evaluation node is constructed, specifically:

[0012] Acquire a preset construction task of the target robotic arm, and acquire a preset working trajectory of the target robotic arm according to the preset construction task;

[0013] Constructing a grid space coordinate system, setting the size of each grid in the grid space coordinate system, and importing the preset working trajectory into the grid space coordinate system;

[0014] Retrieving a preset working trajectory imported into the grid space coordinate system, marking and extracting intersections of the preset working trajectory and grid lines in the grid space coordinate system, and obtaining a plurality of trajectory position evaluation nodes;

[0015] According to the preset construction task, the preset posture information corresponding to each trajectory position evaluation node of the target robot arm is obtained;

[0016] According to the preset posture information corresponding to the target robot arm at each trajectory position evaluation node, a three-dimensional model diagram of the preset posture of the target robot arm at each trajectory position evaluation node is drawn using three-dimensional modeling software.

[0017] Preferably, a blank database is constructed, and each preset posture three-dimensional model image is stored in the blank database to obtain a feature database, specifically:

[0018] Constructing a blank database, and obtaining the number of nodes of the trajectory position evaluation node of the target robotic arm, and dividing a corresponding number of storage spaces in the blank database according to the number of nodes of the trajectory position evaluation node of the target robotic arm;

[0019] Obtaining the evaluation position coordinate information of each trajectory position evaluation node of the target manipulator, and assigning a coordinate coding label to each storage space according to the evaluation position coordinate information of each trajectory position evaluation node;

[0020] The preset three-dimensional model diagram of the target robot arm at each trajectory position evaluation node is obtained, and the preset three-dimensional model diagram of the target robot arm at each trajectory position evaluation node is stored in the storage space after being assigned with coordinate coding labels to obtain a feature database.

[0021] Preferably, the real-time position of the target manipulator is analyzed in combination with the feature database to determine whether the real-time position of the target manipulator is a trajectory position evaluation node. If it is a trajectory position evaluation node, the posture state of the target manipulator at the current position node is further analyzed, specifically:

[0022] During the construction task of the target mechanical arm, the real-time position coordinate information of the target mechanical arm is obtained, and the real-time position coordinate information of the target mechanical arm is imported into the feature database;

[0023] Matching and analyzing the real-time location coordinate information with each coordinate coding tag in the feature database one by one;

[0024] If the real-time position coordinate information does not match the coordinate coding labels in the feature database, the current position node of the target manipulator is marked as a non-track position evaluation node, and the position and posture of the target manipulator are not evaluated at this time;

[0025] If the real-time position coordinate information matches the coordinate coding labels in the feature database, the current position node of the target robot arm is calibrated as a trajectory position evaluation node, and the position and posture of the target robot arm is evaluated.

[0026] Preferably, if the real-time position coordinate information matches the coordinate coding tags in the feature database, the current position node of the target manipulator is calibrated as the trajectory position evaluation node, and the position and posture of the target manipulator are evaluated and processed, specifically:

[0027] If the real-time position coordinate information matches a coordinate coding tag in the feature database, a preset posture three-dimensional model diagram is extracted from the corresponding storage space in the feature database according to the matched coordinate coding tag to obtain a preset posture three-dimensional model diagram of the target manipulator at the current position node;

[0028] Acquire actual posture image information of the target robotic arm, and construct a three-dimensional model diagram of the actual posture of the target robotic arm at the current position node according to the actual posture image information;

[0029] Based on the iterative closest point matching algorithm, the preset pose 3D model image of the target robot arm at the current position node is registered with the actual pose 3D model image;

[0030] After the registration is completed, an axis-aligned bounding box is constructed for the preset pose 3D model image and the actual pose 3D model image of the target robot arm at the current position node, and the vertex coordinates of each vertex in the axis-aligned bounding box are obtained;

[0031] Based on the vertex coordinates of each vertex in the axis-aligned bounding box and using an octree partitioning algorithm, a preset pose three-dimensional model graph of the target robotic arm at the current position node and an actual pose three-dimensional model graph are spatially divided to divide the preset pose three-dimensional model graph of the target robotic arm at the current position node and the actual pose three-dimensional model graph into a plurality of node spaces;

[0032] Perform model analysis on each node space one by one; if a certain node space contains both a preset pose 3D model graph and an actual pose 3D model graph, the node space is defined as an intersection node space;

[0033] If only a preset pose 3D model graph or only an actual pose 3D model graph exists in a certain node space, the node space is defined as a non-intersection node space;

[0034] Calculate the total volume value of all intersection node spaces, and calculate the total volume value of all non-intersection node spaces;

[0035] The total volume value of all intersection node spaces is compared with the total volume value of all non-intersection node spaces, and then multiplied by a preset weight value to obtain an overlap ratio between a preset pose 3D model diagram of the target robot arm at the current position node and an actual pose 3D model diagram;

[0036] If the overlap ratio between the preset pose 3D model diagram of the target robot arm at the current position node and the actual pose 3D model diagram is greater than a preset value, the pose state of the target robot arm at the current position node is marked as a normal state;

[0037] If the overlap ratio between the preset pose 3D model diagram of the target robot arm at the current position node and the actual pose 3D model diagram is not greater than a preset value, the pose state of the target robot arm at the current position node is marked as an offset state.

[0038] Preferably, if the posture state of the target robotic arm at the current position node is an offset state, the target robotic arm is corrected, specifically:

[0039] Obtaining a work diary of the target robotic arm, and obtaining, according to the work diary, posture deviation events that have occurred in the target robotic arm during the working process;

[0040] Obtain the historical correction parameters corresponding to each posture deviation event of the target robot arm, and obtain the correction accuracy of the target robot arm after correction with the corresponding historical correction parameters;

[0041] The historical correction parameters corresponding to the highest correction accuracy are used as the optimal correction parameters of the corresponding posture deviation event, and the optimal correction parameters corresponding to the target manipulator when each posture deviation event occurs are obtained;

[0042] and obtaining historical posture image information of the target manipulator when each posture offset event occurs, and constructing a three-dimensional model diagram of the historical posture corresponding to the target manipulator when each posture offset event occurs according to the corresponding historical posture image information;

[0043] Construct a knowledge graph, and divide the knowledge graph into a plurality of sub-graph nodes, and import the optimal correction parameters and the historical posture three-dimensional model graph corresponding to the target manipulator when each posture deviation event occurs into the corresponding sub-graph nodes;

[0044] If the posture state of the target robot arm at the current position node is an offset state, then obtain the actual posture three-dimensional model diagram of the target robot arm at the current position node;

[0045] Importing the actual posture 3D model graph of the target robotic arm at the current position node into the knowledge graph, and calculating the overlap ratio between the actual posture 3D model graph of the target robotic arm at the current position node and the historical posture 3D model graph in each graph node;

[0046] Compare the overlap ratio between the actual pose 3D model graph of the target robot at the current position node and the historical pose 3D model graph in each atlas node, compare the maximum overlap ratio, and extract the optimal correction parameter in the sub-atlas node corresponding to the maximum overlap ratio;

[0047] The extracted optimal correction parameters are sent to the control terminal so as to perform correction and control processing on the target robot arm through the extracted optimal correction parameters.

[0048] The second aspect of the present invention discloses a large-scale multi-joint hydraulic mechanical arm for building a marine ranch, and a control method for a large-scale multi-joint hydraulic mechanical arm for building a marine ranch applied to any one of the above, comprising:

[0049] The base module is used to fix the robot arm, provide support, and serve as a mounting platform for the hydraulic system and control system;

[0050] The hydraulic system module provides the power required for the movement of the robot arm and controls the movement speed, force and accuracy of the robot arm;

[0051] A robotic arm module, including an arm body, a universal connector and an end effector;

[0052] The control system module controls the movements of the robotic arm, including trajectory, speed, and force;

[0053] The sensor module monitors the working status of the robot arm, including position, speed, force and hydraulic pressure, and feeds back to the control system for adjustment;

[0054] The communication module enables communication between the robotic arm and the onshore control center, as well as the transmission of data and instructions.

[0055] The present invention solves the technical defects existing in the background technology, and the present invention has the following beneficial effects: obtaining the preset working trajectory of the robotic arm, extracting the key trajectory position evaluation nodes, constructing a preset posture three-dimensional model diagram, and storing it in a feature database. In the actual operation process, the system judges the position nodes of the robotic arm through real-time position analysis and matching, and performs posture evaluation. If a posture deviation is found, the system will quickly find the best correction scheme based on the knowledge graph of historical correction parameters and optimal correction parameters, and perform precise correction. Through the present invention, not only can the operating accuracy and efficiency of the robotic arm be significantly improved, but also through historical data and model matching, efficient and accurate correction processing can be achieved, thereby ensuring the smooth progress of the marine ranch construction task, reducing operating errors and resource waste, and improving the stability and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, drawings of other embodiments can be obtained based on these drawings without paying creative work.

[0057] Figure 1 The overall method flow chart of the control method of the robot arm;

[0058] Figure 2 This is a partial method flow chart of the control method of the robot arm;

[0059] Figure 3 is a schematic diagram of the first structure of the robot arm;

[0060] Figure 4 This is a schematic diagram of the second structure of the robot arm. DETAILED DESCRIPTION

[0061] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0062] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0063] The first aspect of the present invention discloses a control method for a large-scale multi-joint hydraulic mechanical arm for building a marine ranch, such as Figure 1 As shown, the following steps are included:

[0064] S102: Acquire a preset working trajectory of the target manipulator, perform feature extraction on the preset working trajectory, obtain a plurality of trajectory position evaluation nodes of the target manipulator, and construct a preset posture three-dimensional model diagram of the target manipulator at each trajectory position evaluation node;

[0065] S104: constructing a blank database, and storing each preset posture three-dimensional model image in the blank database to obtain a feature database;

[0066] S106: performing position analysis on the real-time position of the target robotic arm in combination with the feature database to determine whether the real-time position of the target robotic arm is a trajectory position evaluation node; if it is a trajectory position evaluation node, further analyzing the posture state of the target robotic arm at the current position node;

[0067] S108: If the posture state of the target robotic arm at the current position node is normal, no deviation correction processing is performed on the target robotic arm; if the posture state of the target robotic arm at the current position node is offset, a deviation correction processing is performed on the target robotic arm.

[0068] It should be noted that, first, the preset working trajectory of the target manipulator is obtained and analyzed, the key trajectory position evaluation nodes are extracted, and the preset posture three-dimensional model diagrams of these nodes are constructed. Then, these three-dimensional model diagrams are stored in a blank database to form a feature database. In the actual operation process, the system will use this feature database to analyze the real-time position of the manipulator to determine whether it is located on the preset trajectory position evaluation node. If the manipulator is located on the trajectory position evaluation node, its posture state is further analyzed. If the posture state is normal, no correction processing is required; on the contrary, if the posture state is detected to be offset, correction measures are taken immediately. This method can achieve high-precision path tracking and posture control, thereby improving the construction efficiency and quality in the construction of marine ranches, and reducing the waste of resources, damage and collision risks caused by the deviation of the manipulator movement.

[0069] Preferably, the preset working trajectory of the target manipulator is obtained, and features are extracted for the preset working trajectory to obtain several trajectory position evaluation nodes of the target manipulator, and a preset posture three-dimensional model diagram of the target manipulator at each trajectory position evaluation node is constructed, such as Figure 2 As shown, specifically:

[0070] S202: Acquire a preset construction task of the target robotic arm, and acquire a preset working trajectory of the target robotic arm according to the preset construction task;

[0071] S204: constructing a grid space coordinate system, setting the size of each grid in the grid space coordinate system, and importing the preset working trajectory into the grid space coordinate system;

[0072] S206: searching the preset working trajectory imported into the grid space coordinate system, marking and extracting the intersection points of the preset working trajectory and the grid lines in the grid space coordinate system, and obtaining a plurality of trajectory position evaluation nodes;

[0073] S208: Acquire preset posture information corresponding to each trajectory position evaluation node of the target robot arm according to the preset construction task;

[0074] According to the preset posture information corresponding to the target robot arm at each trajectory position evaluation node, a three-dimensional model diagram of the preset posture of the target robot arm at each trajectory position evaluation node is drawn using three-dimensional modeling software.

[0075] Among them, the "preset construction task" refers to the specific task that the robotic arm needs to complete that is pre-planned and defined, and is formulated in advance by relevant technical personnel.

[0076] It should be noted that the specific tasks that the robot needs to complete are first clarified, and the preset working trajectory of the robot is determined according to the task requirements. A gridded spatial coordinate system is established, and the size of the grid is defined by relevant technicians according to the control accuracy requirements of the robot, which helps to convert the continuous working trajectory into discrete grid points for subsequent processing. The preset working trajectory is imported into the grid spatial coordinate system, and then a series of trajectory position evaluation nodes are marked and extracted by retrieving the intersection of the trajectory and the grid line. These nodes are the key position points in the entire working trajectory. Based on the preset construction task, the posture (i.e., position and posture) information that the robot should have at each trajectory position evaluation node is determined. Using three-dimensional modeling software (such as CAD, UG), according to the preset posture information on each trajectory position evaluation node, a three-dimensional posture model diagram of the robot at these nodes is drawn. Through the above steps, the motion trajectory and posture changes of the robot in the construction of the marine ranch can be accurately planned and simulated. It can ensure that the robot performs tasks according to the predetermined trajectory and posture during the actual operation, thereby improving the accuracy and efficiency of the construction and reducing the occurrence of errors and accidents.

[0077] Preferably, a blank database is constructed, and each preset posture three-dimensional model image is stored in the blank database to obtain a feature database, specifically:

[0078] Constructing a blank database, and obtaining the number of nodes of the trajectory position evaluation node of the target robotic arm, and dividing a corresponding number of storage spaces in the blank database according to the number of nodes of the trajectory position evaluation node of the target robotic arm;

[0079] Obtaining the evaluation position coordinate information of each trajectory position evaluation node of the target manipulator, and assigning a coordinate coding label to each storage space according to the evaluation position coordinate information of each trajectory position evaluation node;

[0080] The preset three-dimensional model diagram of the target robot arm at each trajectory position evaluation node is obtained, and the preset three-dimensional model diagram of the target robot arm at each trajectory position evaluation node is stored in the storage space after being assigned a coordinate coding label to obtain a feature database.

[0081] It should be noted that, first, a blank database is created to store the preset pose three-dimensional model diagram of the robot arm at each trajectory position evaluation node. According to the number of trajectory position evaluation nodes of the robot arm, a corresponding number of storage spaces are divided in the blank database, and each storage space is used to store the data of a trajectory position evaluation node. The evaluation position coordinate information of each trajectory position evaluation node is obtained, and a unique coordinate coding label is assigned to each storage space, which can help quickly locate and identify the data of each node. The preset pose three-dimensional model diagram of the target robot arm at each trajectory position evaluation node is obtained, and these model diagrams are respectively stored in the storage space with the corresponding coordinate coding label. Finally, a complete feature database is formed. Each trajectory position evaluation node has a unique coordinate coding label, so that the data in the database can be efficiently retrieved and accessed, which can effectively improve the response speed and operation efficiency of the system.

[0082] Preferably, the real-time position of the target manipulator is analyzed in combination with the feature database to determine whether the real-time position of the target manipulator is a trajectory position evaluation node. If it is a trajectory position evaluation node, the posture state of the target manipulator at the current position node is further analyzed, specifically:

[0083] During the construction task of the target mechanical arm, the real-time position coordinate information of the target mechanical arm is obtained, and the real-time position coordinate information of the target mechanical arm is imported into the feature database;

[0084] Matching and analyzing the real-time location coordinate information with each coordinate coding tag in the feature database one by one;

[0085] If the real-time position coordinate information does not match the coordinate coding labels in the feature database, the current position node of the target manipulator is marked as a non-track position evaluation node, and the position and posture of the target manipulator are not evaluated at this time;

[0086] If the real-time position coordinate information matches the coordinate coding labels in the feature database, the current position node of the target robot arm is calibrated as a trajectory position evaluation node, and the position and posture of the target robot arm is evaluated.

[0087] It should be noted that in the process of the robotic arm performing the construction task, the position coordinate information of the robotic arm is obtained in real time through the sensor module. The acquired real-time position coordinate information is imported into the feature database. The real-time position coordinate information is matched and analyzed with each coordinate coding label in the feature database one by one. If the real-time position coordinate information does not match all the coordinate coding labels in the feature database, the node of the current robotic arm is calibrated as a non-track position evaluation node. At this time, the posture of the robotic arm is not evaluated. If the real-time position coordinate information matches a certain coordinate coding label in the feature database, the node of the current robotic arm is calibrated as a track position evaluation node. At this time, the posture of the robotic arm needs to be evaluated. When the robotic arm reaches the track position evaluation node, the system will accurately evaluate the posture of the robotic arm, which helps to timely discover and correct the posture deviation of the robotic arm to ensure that it performs the task according to the predetermined trajectory and posture. Through the automated real-time monitoring and evaluation mechanism, the need for manual intervention is reduced, the operation process is simplified, the operation complexity and error probability are reduced, the operation accuracy and efficiency of the robotic arm in the construction of the marine ranch are improved, and the smooth progress of the task is ensured.

[0088] Preferably, if the real-time position coordinate information matches the coordinate coding tags in the feature database, the current position node of the target manipulator is calibrated as the trajectory position evaluation node, and the position and posture of the target manipulator are evaluated and processed, specifically:

[0089] If the real-time position coordinate information matches a coordinate coding tag in the feature database, a preset posture three-dimensional model diagram is extracted from the corresponding storage space in the feature database according to the matched coordinate coding tag to obtain a preset posture three-dimensional model diagram of the target manipulator at the current position node;

[0090] Acquire actual posture image information of the target robotic arm, and construct a three-dimensional model diagram of the actual posture of the target robotic arm at the current position node according to the actual posture image information;

[0091] Based on the iterative closest point matching algorithm, the preset pose 3D model image of the target robot arm at the current position node is registered with the actual pose 3D model image;

[0092] After the registration is completed, an axis-aligned bounding box is constructed for the preset pose 3D model image and the actual pose 3D model image of the target robot arm at the current position node, and the vertex coordinates of each vertex in the axis-aligned bounding box are obtained;

[0093] Based on the vertex coordinates of each vertex in the axis-aligned bounding box and using an octree partitioning algorithm, a preset pose three-dimensional model graph of the target robotic arm at the current position node and an actual pose three-dimensional model graph are spatially divided to divide the preset pose three-dimensional model graph of the target robotic arm at the current position node and the actual pose three-dimensional model graph into a plurality of node spaces;

[0094] Perform model analysis on each node space one by one; if a certain node space contains both a preset pose 3D model graph and an actual pose 3D model graph, the node space is defined as an intersection node space;

[0095] If only a preset pose 3D model graph or only an actual pose 3D model graph exists in a certain node space, the node space is defined as a non-intersection node space;

[0096] Calculate the total volume value of all intersection node spaces, and calculate the total volume value of all non-intersection node spaces;

[0097] The total volume value of all intersection node spaces is ratioed to the total volume value of all non-intersection node spaces and then multiplied by a preset weight value to obtain an overlap ratio between a preset pose 3D model diagram of the target manipulator at the current position node and an actual pose 3D model diagram;

[0098] If the overlap ratio between the preset posture 3D model diagram of the target robot arm at the current position node and the actual posture 3D model diagram is greater than a preset value, the posture state of the target robot arm at the current position node is marked as a normal state;

[0099] If the overlap ratio between the preset pose 3D model diagram of the target robot arm at the current position node and the actual pose 3D model diagram is not greater than a preset value, the pose state of the target robot arm at the current position node is marked as an offset state.

[0100] It should be noted that when the real-time position coordinate information matches a coordinate coding label in the feature database, the preset pose three-dimensional model image is extracted from the storage space corresponding to the matched coordinate coding label in the feature database. The actual pose image information of the target manipulator is obtained, and the actual pose three-dimensional model image is constructed based on these image information. The preset pose three-dimensional model image and the actual pose three-dimensional model image are registered using the iterative closest point matching algorithm (ICP algorithm) to align the two. After the registration is completed, an axis-aligned bounding box is constructed for the preset pose three-dimensional model image and the actual pose three-dimensional model image respectively, and the vertex coordinates of each vertex in the bounding box are obtained. The space of the preset pose three-dimensional model image and the actual pose three-dimensional model image is divided into several node spaces using the octree partitioning algorithm. Each node space is analyzed one by one: if there is both a preset pose three-dimensional model image and an actual pose three-dimensional model image in a certain node space, the node space is defined as an intersection node space. If there is only a preset pose three-dimensional model graph or only an actual pose three-dimensional model graph in a certain node space, the node space is defined as a non-intersection node space. The overlap ratio between the preset pose three-dimensional model graph and the actual pose three-dimensional model graph of the target manipulator at the current position node is obtained according to the total volume value of the intersection node space and the total volume value of all non-intersection node spaces. By calculating the overlap ratio, the evaluation result can be quantified, thereby achieving high-precision pose evaluation. If the overlap ratio is greater than the preset value, the pose state of the target manipulator at the current position node is marked as a normal state. If the overlap ratio is not greater than the preset value, the pose state of the target manipulator at the current position node is marked as an offset state. In general, this step uses the registration processing of the preset pose three-dimensional model graph and the actual pose three-dimensional model graph to accurately evaluate whether the actual pose of the manipulator meets expectations. The ability to monitor the pose state of the manipulator in real time and determine whether it is offset based on the overlap ratio helps to promptly discover and correct the pose deviation of the manipulator and ensure that it performs tasks according to the predetermined trajectory and posture. At the same time, through the calculation of the overlap ratio, the system can intelligently judge the posture state of the robot arm and take corresponding correction measures, which helps to improve the autonomy and reliability of the system, thereby optimizing the operation strategy and path planning of the robot arm and improving overall work efficiency.

[0101] Preferably, if the posture state of the target robotic arm at the current position node is an offset state, the target robotic arm is corrected, specifically:

[0102] Obtaining a work diary of the target robotic arm, and obtaining, according to the work diary, posture deviation events that have occurred in the target robotic arm during the working process;

[0103] Obtain the historical correction parameters corresponding to each posture deviation event of the target robot arm, and obtain the correction accuracy of the target robot arm after correction with the corresponding historical correction parameters;

[0104] The historical correction parameters corresponding to the highest correction accuracy are used as the optimal correction parameters of the corresponding posture deviation event, and the optimal correction parameters corresponding to the target manipulator when each posture deviation event occurs are obtained;

[0105] and obtaining historical posture image information of the target manipulator when each posture offset event occurs, and constructing a three-dimensional model diagram of the historical posture corresponding to the target manipulator when each posture offset event occurs according to the corresponding historical posture image information;

[0106] Construct a knowledge graph, and divide the knowledge graph into a plurality of sub-graph nodes, and import the optimal correction parameters and the historical posture three-dimensional model graph corresponding to the target manipulator when each posture deviation event occurs into the corresponding sub-graph nodes;

[0107] If the posture state of the target robot arm at the current position node is an offset state, then obtain the actual posture three-dimensional model diagram of the target robot arm at the current position node;

[0108] Importing the actual posture 3D model graph of the target robotic arm at the current position node into the knowledge graph, and calculating the overlap ratio between the actual posture 3D model graph of the target robotic arm at the current position node and the historical posture 3D model graph in each graph node;

[0109] Compare the overlap ratio between the actual pose 3D model graph of the target robot at the current position node and the historical pose 3D model graph in each atlas node, compare the maximum overlap ratio, and extract the optimal correction parameter in the sub-atlas node corresponding to the maximum overlap ratio;

[0110] The extracted optimal correction parameters are sent to the control terminal so as to perform correction and control processing on the target robot arm through the extracted optimal correction parameters.

[0111] It should be noted that by obtaining the work diary of the robotic arm, according to the posture offset events in the diary and its historical correction parameters, the optimal correction parameters with the highest correction accuracy are screened out, and the corresponding historical posture three-dimensional model graph is constructed. Then, a knowledge graph is constructed and the optimal correction parameters and the historical posture three-dimensional model graph are imported into the corresponding sub-graph node. When the posture state of the robotic arm at the current position node is an offset state, its actual posture three-dimensional model graph is obtained and imported into the knowledge graph, the overlap ratio between the actual posture three-dimensional model graph and the historical posture three-dimensional model graph in each sub-graph node is calculated, the optimal correction parameters corresponding to the maximum overlap ratio are extracted, and the correction and regulation processing of the robotic arm is performed through the control terminal. This method can achieve efficient and accurate robotic arm correction processing through historical data and model matching, and improve the stability and accuracy of robotic arm operation.

[0112] In summary, the preset working trajectory of the robot arm is obtained, and the key trajectory position evaluation nodes are extracted to construct a preset posture three-dimensional model diagram and store it in the feature database. In the actual operation process, the system determines the position node of the robot arm through real-time position analysis and matching, and performs posture evaluation. If posture deviation is found, the system will quickly find the best correction scheme based on the knowledge graph of historical correction parameters and optimal correction parameters, and perform precise correction. It can not only significantly improve the operating accuracy and efficiency of the robot arm, but also achieve efficient and accurate correction processing through historical data and model matching, thereby ensuring the smooth progress of the marine ranch construction task, reducing operational errors and resource waste, and improving the stability and reliability of the system.

[0113] In addition, during the actual operation of the robot arm, the control method thereof further includes the following steps:

[0114] Based on the big data network, historical fault characteristic data of the target robot arm under various environmental characteristics and continuous working steps are obtained;

[0115] Constructing a Bayesian network model, and taking the various environmental characteristics and the continuous working step length as the first network node, and taking the historical fault characteristic data as the second network node;

[0116] Constructing a Bayesian network topology graph according to the first network node and the second network node, and performing Laplace matrix transformation processing on the Bayesian network topology graph to obtain a Laplace matrix of the Bayesian network topology graph;

[0117] A cosine similarity algorithm is introduced, and the cosine similarity between every two elements in the Laplace matrix is ​​calculated based on the cosine similarity algorithm; if the cosine similarity between two elements of the Laplace matrix is ​​greater than a preset similarity threshold, one of the elements whose cosine similarity is greater than the preset similarity threshold is removed, and so on, the redundancy reduction operation of the Laplace matrix is ​​completed;

[0118] The reduced-redundancy Laplace matrix is ​​introduced into the Bayesian network model for coding learning until the model parameters meet the preset requirements, thereby obtaining a Bayesian network fault prediction model;

[0119] During the construction task of the target robotic arm, the real-time working step and real-time environmental characteristics of the target robotic arm are obtained at a preset time node, and the real-time working step and real-time environmental characteristics are imported into the Bayesian network fault prediction model for prediction to obtain the failure probability of the target robotic arm;

[0120] If the failure probability is greater than a preset failure probability value, the target robotic arm is controlled to stop running and a failure warning message is generated.

[0121] The environmental characteristics include temperature, salinity, wind speed, and electromagnetic interference intensity.

[0122] It should be noted that, based on the big data network, the historical fault feature data of the target manipulator under various environmental characteristics and continuous working steps are obtained. These data contain the fault records of the manipulator under different conditions, providing a basis for subsequent analysis. Various environmental characteristics and continuous working steps are used as the first network node, and the historical fault feature data are used as the second network node. According to the dependency relationship between these nodes, a Bayesian network topology graph is constructed. The Bayesian network topology graph is transformed by Laplace matrix to obtain the Laplace matrix. The similarity between every two elements in the Laplace matrix is ​​calculated by the cosine similarity algorithm. If the cosine similarity of two elements is greater than the preset threshold, one of the elements is removed to reduce redundant information. The reduced Laplace matrix is ​​imported into the Bayesian network model for encoding learning until the model parameters meet the preset requirements. Finally, a Bayesian network fault prediction model is obtained to predict the failure probability of the manipulator.

[0123] During the construction task of the target robotic arm, the real-time working step and real-time environmental characteristics are obtained at the preset time node. These real-time data are imported into the Bayesian network fault prediction model for prediction to obtain the fault probability. If the fault probability is greater than the preset value, the robotic arm is controlled to stop running and a fault warning information is generated. Through big data analysis and Bayesian network models, the failure probability of the robotic arm under different environments and working conditions can be accurately predicted to improve the accuracy of fault prediction. During the operation of the robotic arm, its working status is monitored in real time, and a warning is issued in time when the failure probability exceeds the preset threshold to avoid potential failures and ensure the safe operation of the robotic arm. The fault prediction model is used to optimize the work plan of the robotic arm, arrange work tasks reasonably, and avoid high-risk operations under adverse conditions.

[0124] like Figure 3 , 4 As shown, the second aspect of the present invention discloses a large-scale multi-joint hydraulic mechanical arm for building a marine ranch, which is applied to a control method of a large-scale multi-joint hydraulic mechanical arm for building a marine ranch as described in any one of the items, comprising:

[0125] Base module 1101, used to fix the robot arm, provide support, and serve as a mounting platform for the hydraulic system and control system;

[0126] The hydraulic system module 1102 provides the power required for the movement of the robot arm and controls the movement speed, force and precision of the robot arm;

[0127] The robot arm module includes an arm body 1103, a universal connector 1104 and an end effector; the end effector can be interlocked with the universal connector, and the end effector includes but is not limited to a fixture, a disc cutter, and a chain cutter;

[0128] The control system module controls the movements of the robotic arm, including trajectory, speed, and force;

[0129] The sensor module monitors the working status of the robot arm, including position, speed, force and hydraulic pressure, and feeds back to the control system for adjustment;

[0130] The communication module enables communication between the robotic arm and the onshore control center, as well as the transmission of data and instructions.

[0131] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0132] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0133] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0134] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0135] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0136] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A control method for a large-scale multi-joint hydraulic mechanical arm for building a marine ranch, characterized in that: The following steps are involved: Obtaining a preset working trajectory of the target robotic arm, performing feature extraction on the preset working trajectory, obtaining a number of trajectory position evaluation nodes of the target robotic arm, and constructing a preset posture three-dimensional model diagram of the target robotic arm at each trajectory position evaluation node; Constructing a blank database, and storing each preset posture three-dimensional model image in the blank database to obtain a feature database; Performing a position analysis on the real-time position of the target robotic arm in combination with the feature database to determine whether the real-time position of the target robotic arm is a trajectory position evaluation node; if it is a trajectory position evaluation node, further analyzing the posture state of the target robotic arm at the current position node; If the posture state of the target robot arm at the current position node is normal, no deviation correction processing will be performed on the target robot arm; If the posture state of the target robot arm at the current position node is an offset state, the target robot arm is corrected.

2. The control method of a large-scale multi-joint hydraulic mechanical arm for building a marine ranch according to claim 1 is characterized in that: The preset working trajectory of the target manipulator is obtained, and features are extracted from the preset working trajectory to obtain several trajectory position evaluation nodes of the target manipulator, and a preset posture 3D model diagram of the target manipulator at each trajectory position evaluation node is constructed, specifically: Acquire a preset construction task of the target robotic arm, and acquire a preset working trajectory of the target robotic arm according to the preset construction task; Constructing a grid space coordinate system, setting the size of each grid in the grid space coordinate system, and importing the preset working trajectory into the grid space coordinate system; Retrieving a preset working trajectory imported into the grid space coordinate system, marking and extracting intersections of the preset working trajectory and grid lines in the grid space coordinate system, and obtaining a plurality of trajectory position evaluation nodes; According to the preset construction task, the preset posture information corresponding to each trajectory position evaluation node of the target robot arm is obtained; According to the preset posture information corresponding to the target robot arm at each trajectory position evaluation node, a three-dimensional model diagram of the preset posture of the target robot arm at each trajectory position evaluation node is drawn using three-dimensional modeling software.

3. The control method of a large-scale multi-joint hydraulic mechanical arm for building a marine ranch according to claim 1 is characterized in that: A blank database is constructed, and each preset posture three-dimensional model image is stored in the blank database to obtain a feature database, specifically: Constructing a blank database, and obtaining the number of nodes of the trajectory position evaluation node of the target robotic arm, and dividing a corresponding number of storage spaces in the blank database according to the number of nodes of the trajectory position evaluation node of the target robotic arm; Obtaining the evaluation position coordinate information of each trajectory position evaluation node of the target manipulator, and assigning a coordinate coding label to each storage space according to the evaluation position coordinate information of each trajectory position evaluation node; The preset three-dimensional model diagram of the target robot arm at each trajectory position evaluation node is obtained, and the preset three-dimensional model diagram of the target robot arm at each trajectory position evaluation node is stored in the storage space after being assigned with coordinate coding labels to obtain a feature database.

4. The control method of a large-scale multi-joint hydraulic mechanical arm for building a marine ranch according to claim 1 is characterized in that: Combined with the feature database, the real-time position of the target manipulator is analyzed to determine whether the real-time position of the target manipulator is a trajectory position evaluation node. If it is a trajectory position evaluation node, the posture state of the target manipulator at the current position node is further analyzed, specifically: During the construction task of the target mechanical arm, the real-time position coordinate information of the target mechanical arm is obtained, and the real-time position coordinate information of the target mechanical arm is imported into the feature database; Matching and analyzing the real-time location coordinate information with each coordinate coding tag in the feature database one by one; If the real-time position coordinate information does not match the coordinate coding labels in the feature database, the current position node of the target manipulator is marked as a non-track position evaluation node, and the position and posture of the target manipulator are not evaluated at this time; If the real-time position coordinate information matches the coordinate coding labels in the feature database, the current position node of the target robot arm is calibrated as a trajectory position evaluation node, and the position and posture of the target robot arm is evaluated.

5. The control method of a large-scale multi-joint hydraulic mechanical arm for building a marine ranch according to claim 4 is characterized in that: If the real-time position coordinate information matches the coordinate coding labels in the feature database, the current position node of the target manipulator is marked as the trajectory position evaluation node. At this time, the position and posture of the target manipulator are evaluated, specifically: If the real-time position coordinate information matches a coordinate coding tag in the feature database, a preset posture three-dimensional model diagram is extracted from the corresponding storage space in the feature database according to the matched coordinate coding tag to obtain a preset posture three-dimensional model diagram of the target manipulator at the current position node; Acquire actual posture image information of the target robotic arm, and construct a three-dimensional model diagram of the actual posture of the target robotic arm at the current position node according to the actual posture image information; Based on the iterative closest point matching algorithm, the preset pose 3D model image of the target robot arm at the current position node is registered with the actual pose 3D model image; After the registration is completed, an axis-aligned bounding box is constructed for the preset pose 3D model image and the actual pose 3D model image of the target robot arm at the current position node, and the vertex coordinates of each vertex in the axis-aligned bounding box are obtained; Based on the vertex coordinates of each vertex in the axis-aligned bounding box and using an octree partitioning algorithm, a preset pose three-dimensional model graph of the target robotic arm at the current position node and an actual pose three-dimensional model graph are spatially divided to divide the preset pose three-dimensional model graph of the target robotic arm at the current position node and the actual pose three-dimensional model graph into a plurality of node spaces; Perform model analysis on each node space one by one; if a certain node space contains both a preset pose 3D model graph and an actual pose 3D model graph, the node space is defined as an intersection node space; If only a preset pose 3D model graph or only an actual pose 3D model graph exists in a certain node space, the node space is defined as a non-intersection node space; Calculate the total volume value of all intersection node spaces, and calculate the total volume value of all non-intersection node spaces; The total volume value of all intersection node spaces is compared with the total volume value of all non-intersection node spaces, and then multiplied by a preset weight value to obtain an overlap ratio between a preset pose 3D model diagram of the target robot arm at the current position node and an actual pose 3D model diagram; If the overlap ratio between the preset pose 3D model diagram of the target robot arm at the current position node and the actual pose 3D model diagram is greater than a preset value, the pose state of the target robot arm at the current position node is marked as a normal state; If the overlap ratio between the preset pose 3D model diagram of the target robot arm at the current position node and the actual pose 3D model diagram is not greater than a preset value, the pose state of the target robot arm at the current position node is marked as an offset state.

6. The control method of a large-scale multi-joint hydraulic mechanical arm for building a marine ranch according to claim 5 is characterized in that: If the target robot arm is in an offset state at the current position node, the target robot arm is corrected as follows: Obtaining a work diary of the target robotic arm, and obtaining, according to the work diary, posture deviation events that have occurred in the target robotic arm during the working process; Obtain the historical correction parameters corresponding to each posture deviation event of the target robot arm, and obtain the correction accuracy of the target robot arm after correction with the corresponding historical correction parameters; The historical correction parameters corresponding to the highest correction accuracy are used as the optimal correction parameters of the corresponding posture deviation event, and the optimal correction parameters corresponding to the target manipulator when each posture deviation event occurs are obtained; and obtaining historical posture image information of the target manipulator when each posture offset event occurs, and constructing a three-dimensional model diagram of the historical posture corresponding to the target manipulator when each posture offset event occurs according to the corresponding historical posture image information; Construct a knowledge graph, and divide the knowledge graph into a plurality of sub-graph nodes, and import the optimal correction parameters and the historical posture three-dimensional model graph corresponding to the target manipulator when each posture deviation event occurs into the corresponding sub-graph nodes; If the posture state of the target robot arm at the current position node is an offset state, then obtain the actual posture three-dimensional model diagram of the target robot arm at the current position node; Importing the actual posture 3D model graph of the target robotic arm at the current position node into the knowledge graph, and calculating the overlap ratio between the actual posture 3D model graph of the target robotic arm at the current position node and the historical posture 3D model graph in each graph node; Compare the overlap ratio between the actual pose 3D model graph of the target robot at the current position node and the historical pose 3D model graph in each atlas node, compare the maximum overlap ratio, and extract the optimal correction parameter in the sub-atlas node corresponding to the maximum overlap ratio; The extracted optimal correction parameters are sent to the control terminal so as to perform correction and control processing on the target robot arm through the extracted optimal correction parameters.

7. A large-scale multi-joint hydraulic mechanical arm for building a marine ranch, applied to a control method of a large-scale multi-joint hydraulic mechanical arm for building a marine ranch as claimed in any one of claims 1 to 6, characterized in that: include: The base module is used to fix the robot arm, provide support, and serve as a mounting platform for the hydraulic system and control system; The hydraulic system module provides the power required for the movement of the robot arm and controls the movement speed, force and accuracy of the robot arm; A robotic arm module, including an arm body, a universal connector and an end effector; The control system module controls the movements of the robotic arm, including trajectory, speed, and force; The sensor module monitors the working status of the robot arm, including position, speed, force and hydraulic pressure, and feeds back to the control system for adjustment; The communication module enables communication between the robotic arm and the onshore control center, as well as the transmission of data and instructions.

Images