An adaptive control system and method for underwater robot based on data analysis

Abstract

The invention discloses an underwater robot adaptive control system and method based on data analysis, which relates to the field of automatic control technology. The invention collects environmental data and underwater robot data during the operation of the underwater robot, constructs an operation data set S, performs feature extraction, obtains the acceleration fluctuation coefficient and the attitude angular velocity fluctuation coefficient, and summarizes and calculates to obtain a stability score WD, evaluates the operation state of the underwater robot, and predicts the water flow data at a future time point t+k when the operation state of the underwater robot is abnormal. The robot dynamics model is combined to obtain the underwater robot's adjustment thrust, and a genetic algorithm is used to obtain the underwater robot's optimal thrust distribution plan, so as to control the underwater robot to perform the operation task, effectively cope with strong turbulence and water flow reversal, reduce the risk of operation failure and damage of the underwater robot, and improve the operation stability and safety of the underwater robot.

Description

Technical Field

[0001] The present invention relates to the field of automation control technology, and in particular to an underwater robot adaptive control system and method based on data analysis. Background Art

[0002] The research and application of adaptive control systems of underwater robots belong to the fields of automatic control technology, robotics and marine engineering. Automatic control technology has developed rapidly in recent years and has been gradually applied to the control systems of underwater robots. Underwater robots are mainly used for tasks such as seabed exploration, resource development and environmental monitoring. With the increasing application of underwater robots in deep-sea exploration, seabed resource detection and marine environmental monitoring, the environment faced by robots has become more complex and unpredictable. In these tasks, robots often need to deal with extremely complex water flow environments, especially sudden changes such as turbulence and water flow reversal. Therefore, it is particularly important to research and develop control systems that can cope with these dynamic water flow changes.

[0003] In the Chinese invention patent application with publication number CN117055585A, an intelligent control method and system for an intelligent underwater robot is disclosed. The method includes: creating a three-dimensional underwater space grid map through a sensor to determine the initial position and target position of the underwater robot; dividing the three-dimensional underwater space grid map into a number of underwater space movement layers according to the initial position and target position, and determining the target direction movement according to the underwater space obstacle detection results in each of the underwater space movement layers, and the target direction movement includes lateral movement and transverse movement; in the process of moving along the target direction, calculating the actual distance difference, adjusting the standard movement speed of the underwater robot according to the actual distance difference and the preset standard distance difference, and moving to the target position according to the adjusted movement speed, the invention improves the accuracy of controlling the posture of the underwater robot.

[0004] The above intelligent control method and system of an intelligent underwater robot can control the target direction of the underwater robot and adjust the posture of the underwater robot through the lateral detection results and the longitudinal detection results when the robot has a direction deviation when moving underwater.

[0005] However, although the intelligent control method and system of this intelligent underwater robot can control the underwater robot and adjust the path in real time, it lacks feedback on sudden changes in water flow. The sudden change in water flow will cause the robot to have a delayed response during the control process, making it difficult to adapt to the sudden increase in flow rate or reverse flow rate, causing the robot to lose control or deviate from the predetermined course. In addition, traditional control methods are difficult to optimize energy consumption. When encountering violent water flow, the robot needs to consume more energy to maintain a stable state of motion, which in turn affects the success rate of the mission and the robot's operating efficiency.

[0006] To this end, the present invention provides an underwater robot adaptive control system and method based on data analysis. Summary of the invention

[0007] 1. Technical issues to be resolved

[0008] In view of the shortcomings of the prior art, the present invention provides an underwater robot adaptive control system and method based on data analysis. By integrating an intelligent sensor group and a deep learning algorithm, the operation stability and efficiency of the underwater robot are improved. First, the environmental data and robot data during the operation of the underwater robot are collected in real time to establish an operation data set S to provide comprehensive basic data for subsequent analysis. The stability evaluation module is combined with the acceleration fluctuation coefficient to and attitude angular rate fluctuation coefficient , accurately judge the robot's operating status, discover abnormalities in time and start the water flow change prediction module to ensure the safe operation of the robot in a complex environment, and in the thrust optimization module, obtain the optimal thrust distribution plan through the combination of the robot dynamics model and the genetic algorithm, further improve the robot's operating efficiency, and finally, generate control instructions to ensure that the robot can perform tasks according to the optimal thrust distribution plan, achieve the goals of energy saving, high efficiency and stable operation, and solve the problems in the above-mentioned background technology.

[0009] (II) Technical solution

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions: an underwater robot adaptive control system based on data analysis, including a data acquisition module, a stability evaluation module, a water flow change prediction module, a thrust optimization module and an intelligent control module;

[0011] The data acquisition module is used to deploy an intelligent sensor group on the underwater robot to collect environmental data and underwater robot data during the operation of the underwater robot in real time and construct an operation data set S;

[0012] The stability evaluation module is used to obtain the acceleration fluctuation coefficient according to the operation data set S. and attitude angular rate fluctuation coefficient , and perform summary calculation to obtain the stability score WD of the underwater robot, and preset the stability threshold WDYZ, compare and analyze the stability score WD and the stability threshold WDYZ, and evaluate the operating status of the underwater robot. If the stability score WD is greater than or equal to the stability threshold WDYZ, the operating status of the underwater robot is determined to be abnormal, and the water flow change prediction module is started;

[0013] The water flow change prediction module is used to collect historical environmental data and historical underwater robot data during the operation of the underwater robot, and use the long short-term memory network LSTM algorithm to build a water flow change prediction model, combine the historical environmental data, obtain the water flow data at the future time point t+k, and build a water flow data time series set G;

[0014] The thrust optimization module is used to obtain the target thrust of the underwater robot using the robot dynamics model based on the operation data set S and the water flow data time series set G in combination with the underwater robot operation task. and compensating thrust , and according to the target thrust of the underwater robot and compensating thrust , obtain the underwater robot to adjust the thrust , and use genetic algorithm to obtain the optimal thrust distribution scheme of the underwater robot;

[0015] The intelligent control module is used to generate control instructions according to the optimal thrust distribution plan, control the underwater robot to distribute thrust, and perform operating tasks.

[0016] Preferably, the data acquisition module is used to deploy an intelligent sensor group on the underwater robot to sense the water flow changes in the underwater robot's operating environment and collect environmental data and underwater robot data during the underwater robot's operation in real time, wherein the intelligent sensor group includes a water flow sensor, a turbulence detector, a high-precision GPS and an inertial measurement unit IMU;

[0017] The environmental data includes a water velocity vector , Water flow direction and turbulence intensity , where the water velocity vector The specific manifestations are:

[0018] ;

[0019] in, , and Respectively represent the flow rate of water in the x-axis, y-axis and z-axis;

[0020] The underwater robot data includes the acceleration feature vector of the underwater robot , underwater robot velocity vector , position vector , Pitch angle , Roll Angle and yaw angle , where the acceleration eigenvector and the underwater robot velocity vector The specific manifestations are:

[0021] ;

[0022] ;

[0023] in, , and Respectively represent the acceleration of the underwater robot on the x-axis, y-axis and z-axis, Respectively represent the movement speed of the underwater robot on the x-axis, y-axis and z-axis;

[0024] Based on the acquired environmental data and underwater robot data, an operation data set S is constructed.

[0025] Preferably, the stability assessment module includes a feature extraction unit and an anomaly assessment unit;

[0026] The feature extraction unit is used to perform summary calculation based on the operation data set S to obtain the acceleration fluctuation coefficient and attitude angular rate fluctuation coefficient ;

[0027] The acceleration fluctuation coefficient The method of obtaining is:

[0028] ;

[0029] In the formula, , and Respectively indicate time points The acceleration of the underwater robot on the x-axis, y-axis and z-axis is , and They represent the mean acceleration of the underwater robot on the x-axis, y-axis, and z-axis, respectively, n represents the total time point of data collection, i={1, 2, 3, ..., n};

[0030] The attitude angular rate fluctuation coefficient The method of obtaining is:

[0031] ;

[0032] In the formula, , and Respectively represent the pitch angle , Roll Angle and yaw angle The weight coefficient of , and Respectively represent the pitch angle , Roll Angle and yaw angle The standard deviation of the angular velocity.

[0033] Preferably, the abnormality assessment unit is used to and attitude angular rate fluctuation coefficient , perform summary calculation to obtain the stability score WD of the underwater robot. The stability score WD is obtained as follows:

[0034] ;

[0035] In the formula, and Denote acceleration fluctuation coefficients and the weight coefficient of the attitude angular rate fluctuation coefficient ZT, represents the first correction constant;

[0036] A stability threshold WDYZ is preset, and the stability threshold WDYZ and the stability score WD are compared and analyzed to evaluate the operating status of the underwater robot. If the stability score WD is less than the stability threshold WDYZ, the operating status of the underwater robot is determined to be stable and no processing is required. If the stability score WD is greater than or equal to the stability threshold WDYZ, the operating status of the underwater robot is determined to be abnormal. At this time, the water flow change prediction module is started to adjust the operation of the underwater robot.

[0037] Preferably, the water flow change prediction module is used to construct a water flow change prediction model based on the long short-term memory network LSTM algorithm, and use the water flow change prediction model to predict and obtain water flow data at a future time point t+k, wherein the water flow data includes water flow velocity , Water flow direction and turbulence intensity ;

[0038] According to the underwater robot operation database, historical environmental data and historical underwater robot data during the underwater robot operation process are obtained to construct a multidimensional data vector H. The multidimensional data vector H is specifically expressed in the form of:

[0039] ;

[0040] The multidimensional data vector H is input into the water flow change prediction model for model training, and the mean square error method MSE is used to optimize the parameters of the water flow change prediction model;

[0041] Using the trained water flow change prediction model, the environmental data collected in real time during the operation of the underwater robot is input into the water flow change prediction model to obtain the water flow data at the future time point t+k, and construct a water flow data time series set G, which includes the water flow velocity feature vector , Water flow direction and turbulence intensity .

[0042] Preferably, the thrust optimization module includes a thrust calculation unit, an objective function construction unit and a thrust allocation unit;

[0043] The thrust calculation unit is used to obtain the target thrust of the underwater robot using the robot dynamics model based on the operation data set S and the water flow data time series set G, in combination with the preset underwater robot operation task. and compensating thrust ,Among these, the underwater robot operation task refers to reach the target location and maintain the current operation speed;

[0044] The target thrust The method of obtaining is:

[0045] ;

[0046] In the formula, represents the mass matrix of the underwater robot, represents the target acceleration eigenvector of the underwater robot, represents the control matrix, represents the velocity vector of the underwater robot, represents the resistance matrix, Represents the current position vector of the underwater robot;

[0047] The compensating thrust The method of obtaining is:

[0048] ;

[0049] In the formula, represents the water flow influence coefficient, Indicates a future time point The water velocity vector, represents the velocity vector of the underwater robot;

[0050] Push the target and compensating thrust Add together to get the underwater robot's adjusted thrust .

[0051] Preferably, the objective function construction unit is used to construct an objective function J(f) for the purpose of minimizing energy consumption and to set up constraints. The objective function J(f) is specifically expressed as follows:

[0052] ;

[0053] in, represents the total number of thrusters of the underwater robot, represents the propulsion efficiency matrix of the underwater robot, The underwater robot The thrust vector of each thruster, The underwater robot The transpose of the thrust vector of each thruster, The underwater robot Energy consumption of each thruster, q={1, 2, 3, ..., m};

[0054] The propulsion efficiency matrix of the underwater robot Obtained through the underwater robot manufacturer database;

[0055] The constraint condition refers to the thrust constraint, that is, the thrust vector of each thruster ≤Maximum thrust vector .

[0056] Preferably, the thrust distribution unit is used to obtain an optimal thrust distribution scheme based on a genetic algorithm. The specific process of obtaining the optimal thrust distribution scheme is as follows:

[0057] Adjust the thrust according to the underwater robot , randomly generate g thrust allocation schemes, and for each thrust allocation scheme, obtain the objective function value of each thrust allocation scheme ;

[0058] And according to the objective function value of each thrust distribution scheme , calculate the probability of selecting the thrust allocation scheme , where the probability of selecting the thrust allocation scheme is The calculation method is:

[0059] ;

[0060] In the formula, represents the probability of selecting the jth thrust allocation scheme, represents the objective function value of the j-th thrust distribution scheme, represents the total objective function value of all thrust distribution schemes, ;

[0061] Based on the probability of selecting the thrust distribution scheme obtained , using the roulette wheel selection method, the probability of selecting all thrust allocation schemes Map to the range {0, 1} and generate a random number r, where r∈{0, 1}, and accumulate the probability of selection of each thrust allocation scheme , until the probability of selecting the thrust allocation scheme is Greater than or equal to the random number r, output the probability of selection The thrust allocation scheme is greater than or equal to the random number r, and a crossover operation is performed, wherein the crossover operation refers to the thrust allocation strategy of randomly exchanging two thrust allocation schemes using a single-point exchange method, including the thrust sizes of different thrusters;

[0062] Calculate the objective function value of the thrust distribution scheme after the crossover operation , and the objective function value of the thrust distribution scheme after the crossover operation The objective function value of the thrust distribution scheme before the crossover operation For comparison, if the objective function value of the thrust distribution scheme after the cross operation is Greater than the objective function value of the thrust distribution scheme before the crossover operation , the thrust allocation scheme after the crossover operation is retained to obtain multiple thrust allocation schemes, and the thrust allocation scheme is iteratively updated repeatedly until the objective function value of the thrust allocation scheme no longer changes. At this time, the optimal thrust allocation scheme is output.

[0063] Preferably, the intelligent control module is used to transmit the optimal thrust distribution scheme to the underwater robot control system, generate thrust distribution instructions, and transmit the thrust distribution instructions to the underwater robot to control the underwater robot to perform underwater operations.

[0064] An underwater robot adaptive control method based on data analysis comprises the following steps:

[0065] Step 1: deploy an intelligent sensor group on the underwater robot to collect environmental data and underwater robot data during the operation of the underwater robot in real time, and construct an operation data set S;

[0066] Step 2: Obtain the acceleration fluctuation coefficient based on the operation data set S and attitude angular rate fluctuation coefficient , and perform summary calculation to obtain the stability score WD of the underwater robot, and preset the stability threshold WDYZ, compare and analyze the stability score WD and the stability threshold WDYZ, and evaluate the operating status of the underwater robot. If the stability score WD is greater than or equal to the stability threshold WDYZ, the operating status of the underwater robot is determined to be abnormal, and the water flow change prediction module is started;

[0067] Step 3: Collect historical environmental data and historical underwater robot data during the underwater robot operation process, and use the long short-term memory network LSTM algorithm to build a water flow change prediction model. Combined with the historical environmental data, obtain the water flow data at the future time point t+k, and build a water flow data time series set G;

[0068] Step 4: Based on the operation data set S and the water flow data time series set G, combined with the underwater robot operation task, use the robot dynamics model to obtain the target thrust of the underwater robot and compensating thrust , and according to the target thrust of the underwater robot and compensating thrust , obtain the underwater robot to adjust the thrust , and use genetic algorithm to obtain the optimal thrust distribution scheme of the underwater robot;

[0069] Step 5: Generate control instructions based on the optimal thrust distribution plan to control the underwater robot to distribute thrust and perform the operation task.

[0070] The present invention provides an underwater robot adaptive control system and method based on data analysis, which has the following beneficial effects:

[0071] (1) The stability assessment module can effectively prevent the risk of the robot losing control in a complex underwater environment by monitoring the underwater robot’s operating status in real time. and attitude angular rate fluctuation coefficient The calculation can evaluate the stability score WD of the underwater robot in real time and compare it with the preset stability threshold WDYZ. If it is found that the stability score WD of the underwater robot is greater than or equal to the stability threshold WDYZ, the system will promptly start the water flow change prediction module to ensure that the robot can make adjustments according to environmental changes. This mechanism can effectively respond to emergencies such as strong turbulence or water flow reversal, reduce the risk of operation failure or damage of the underwater robot due to environmental changes, and thus improve the operation stability and safety of the underwater robot.

[0072] (2) Through the combination of thrust optimization module and genetic algorithm, the system can accurately calculate and optimize the optimal thrust distribution plan of the underwater robot. This module uses the robot dynamics model and the water flow data time series set G, combined with the underwater robot's operating tasks, such as maintaining the target speed or reaching a specific position. On the basis of considering the water flow changes and the state of the underwater robot, the thrust distribution plan is optimized. The use of genetic algorithm to optimize the thrust distribution plan can minimize energy consumption and improve operating efficiency. By effectively allocating the thrust of each thruster, the robot can maintain efficient work in various complex environments, extend the operating time and save energy, and reduce unnecessary energy waste.

[0073] (3) Through the water flow change prediction module, combined with the long short-term memory network (LSTM) algorithm, the system can predict future water flow changes. The system trains the prediction model based on historical environmental data and underwater robot data. It can predict future water flow changes in advance and adjust the robot's thrust distribution according to the prediction results. The introduction of this module enables the system to have forward-looking and adaptive capabilities. It can effectively cope with complex dynamic water flow environments and automatically adjust the robot's motion trajectory and thrust distribution, thereby improving the success rate of the operation task. When facing uncertain environmental conditions, the robot can quickly adjust its operating strategy to ensure the continuity and stability of the operation and adapt to changes in the underwater environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 This is a block diagram of an underwater robot adaptive control system based on data analysis in the present invention.

[0075] Figure 2 The present invention is a flow chart of an underwater robot adaptive control method based on data analysis. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] Example 1

[0078] See also Figure 1 , the present invention provides an underwater robot adaptive control system based on data analysis, including a data acquisition module, a stability evaluation module, a water flow change prediction module, a thrust optimization module and an intelligent control module;

[0079] The data acquisition module is used to deploy an intelligent sensor group on the underwater robot to collect environmental data and underwater robot data during the operation of the underwater robot in real time and construct an operation data set S;

[0080] The stability evaluation module is used to obtain the acceleration fluctuation coefficient according to the operation data set S. and attitude angular rate fluctuation coefficient , and perform summary calculation to obtain the stability score WD of the underwater robot, and preset the stability threshold WDYZ, compare and analyze the stability score WD and the stability threshold WDYZ, and evaluate the operating status of the underwater robot. If the stability score WD is greater than or equal to the stability threshold WDYZ, the operating status of the underwater robot is determined to be abnormal, and the water flow change prediction module is started;

[0081] The water flow change prediction module is used to collect historical environmental data and historical underwater robot data during the operation of the underwater robot, and use the long short-term memory network LSTM algorithm to build a water flow change prediction model, combine the historical environmental data, obtain the water flow data at the future time point t+k, and build a water flow data time series set G;

[0082] The thrust optimization module is used to obtain the target thrust of the underwater robot using the robot dynamics model based on the operation data set S and the water flow data time series set G in combination with the underwater robot operation task. and compensating thrust , and according to the target thrust of the underwater robot and compensating thrust , obtain the underwater robot to adjust the thrust , and use genetic algorithm to obtain the optimal thrust distribution scheme of the underwater robot;

[0083] The intelligent control module is used to generate control instructions according to the optimal thrust distribution plan, control the underwater robot to distribute thrust, and perform operating tasks.

[0084] In the embodiment, by integrating multiple modules, the operation efficiency and stability of the underwater robot in a complex environment are significantly improved. First, the data acquisition module collects environmental data and robot data in real time by deploying intelligent sensors to ensure that the system can perceive the changes in the underwater operation environment in real time. Secondly, the stability evaluation module calculates the stability score WD of the underwater robot based on the collected operation data, timely evaluates the operation status of the robot and starts the water flow change prediction when the stability is abnormal, to ensure that the robot does not lose control or malfunction in a complex flow environment. The water flow change prediction module accurately predicts future water flow changes based on historical data and LSTM models, prepares for operations in advance, and avoids the robot being affected by unforeseen water flow mutations. In addition, the thrust optimization module combines the robot dynamics model and water flow data, uses genetic algorithms to optimize thrust distribution, reduces energy consumption, and ensures that the robot can complete the task efficiently. Finally, the intelligent control module accurately controls the robot operation according to the optimized thrust distribution scheme, improves operation stability and efficiency, and reduces energy waste. Overall, through real-time data analysis, dynamic optimization of thrust distribution and intelligent decision-making, the autonomy, energy efficiency and environmental adaptability of the underwater robot are effectively improved, and it can perform efficient and safe tasks in complex and changing underwater environments.

[0085] Example 2

[0086] Please refer to Figure 1 Specifically: the data acquisition module is used to deploy an intelligent sensor group on the underwater robot to sense the water flow changes in the underwater robot's operating environment and collect environmental data and underwater robot data during the underwater robot's operation in real time, wherein the intelligent sensor group includes a water flow sensor, a turbulence detector, a high-precision GPS and an inertial measurement unit IMU;

[0087] The environmental data includes a water velocity vector , Water flow direction and turbulence intensity , where the water velocity vector The specific manifestations are:

[0088] ;

[0089] in, , and Respectively represent the flow rate of water in the x-axis, y-axis and z-axis;

[0090] The underwater robot data includes the acceleration feature vector of the underwater robot , underwater robot velocity vector , position vector , Pitch angle , Roll Angle and yaw angle , where the acceleration eigenvector and the underwater robot velocity vector The specific manifestations are:

[0091] ;

[0092] ;

[0093] in, , and Respectively represent the acceleration of the underwater robot on the x-axis, y-axis and z-axis, Respectively represent the movement speed of the underwater robot on the x-axis, y-axis and z-axis;

[0094] The water velocity vector and water flow direction Obtained through water flow sensor;

[0095] The turbulence intensity Obtained through turbulence sounders;

[0096] The acceleration characteristic vector of the underwater robot is , underwater robot velocity vector , Pitch angle , Roll Angle and yaw angle Obtained through the inertial measurement unit IMU;

[0097] The position vector Acquired through high-precision GPS;

[0098] Based on the acquired environmental data and underwater robot data, an operation data set S is constructed.

[0099] In the embodiment, by deploying an intelligent sensor group on the underwater robot, the data acquisition module can perceive the changes in the underwater operating environment in real time, and provide the system with comprehensive and accurate environmental data and robot status data. These data come from multiple high-precision sensors, including water flow sensors, turbulence detectors, high-precision GPS and inertial measurement devices IMU. The water flow sensor can provide water flow velocity vector and direction data, the turbulence detector obtains the turbulence intensity in the water flow, and the IMU sensor provides the robot's acceleration characteristics, velocity vector and posture data, such as pitch angle, roll angle and yaw angle. The high-precision GPS is used to obtain the precise position of the robot. These environmental and robot data are combined to form an operation data set S, which serves as the input basis for subsequent modules and can accurately reflect the real-time operation status and environmental changes of the underwater robot, ensure the reliability of subsequent analysis and control decisions, and enable the system to fully perceive the complex underwater operating environment, especially in extreme cases such as turbulence or water flow reversal, to provide data support for stability evaluation and water flow prediction, and effectively improve the autonomous control capability of the underwater robot, optimize thrust distribution, reduce energy waste, and enhance the stability and safety of task execution.

[0100] Example 3

[0101] Please refer to Figure 1 ,Specifically: the stability evaluation module includes a feature extraction unit and an omaly evaluation unit;

[0102] The feature extraction unit is used to perform summary calculation based on the operation data set S to obtain the acceleration fluctuation coefficient and attitude angular rate fluctuation coefficient ;

[0103] The acceleration fluctuation coefficient The method of obtaining is:

[0104] ;

[0105] In the formula, , and Respectively indicate time points The acceleration of the underwater robot on the x-axis, y-axis and z-axis is , and They represent the mean acceleration of the underwater robot on the x-axis, y-axis, and z-axis, respectively, n represents the total time point of data collection, i={1, 2, 3, ..., n};

[0106] The attitude angular rate fluctuation coefficient The method of obtaining is:

[0107] ;

[0108] In the formula, , and Respectively represent the pitch angle , Roll Angle and yaw angle The weight coefficient of , and Respectively represent the pitch angle , Roll Angle and yaw angle The specific value of the weight coefficient is set by the customer according to the actual situation. , , ,and .

[0109] The abnormality assessment unit is used to determine the acceleration fluctuation coefficient according to the acceleration fluctuation coefficient. and attitude angular rate fluctuation coefficient , perform summary calculation to obtain the stability score WD of the underwater robot. The stability score WD is obtained as follows:

[0110] ;

[0111] In the formula, and Denote acceleration fluctuation coefficients and the weight coefficient of the attitude angular rate fluctuation coefficient ZT, Represents the first correction constant, where the specific value of the weight coefficient is set by the customer according to the actual situation. , ,and ;

[0112] A stability threshold WDYZ is preset, and the stability threshold WDYZ and the stability score WD are compared and analyzed to evaluate the operating status of the underwater robot. If the stability score WD is less than the stability threshold WDYZ, the operating status of the underwater robot is determined to be stable and no processing is required. If the stability score WD is greater than or equal to the stability threshold WDYZ, the operating status of the underwater robot is determined to be abnormal. At this time, the water flow change prediction module is started to adjust the operation of the underwater robot.

[0113] In the embodiment, by introducing the stability evaluation module, the system realizes real-time monitoring and dynamic evaluation of the stability of the underwater robot during operation. The module calculates the acceleration fluctuation coefficient through the feature extraction unit. and attitude angular rate fluctuation coefficient ZT, thereby effectively quantifying the stability performance of the underwater robot in different operating environments. Through real-time monitoring of the underwater robot's acceleration and attitude angular rate, abnormal fluctuations can be discovered in time, reflecting the risk of unstable operation that the robot may face. In addition, the abnormal assessment unit calculates the stability score WD of the underwater robot based on the extracted features, and compares it with the preset stability threshold WDYZ. If the stability score WD is greater than or equal to the stability threshold WDYZ, the system determines that the underwater robot's operating state is abnormal and automatically starts the water flow change prediction module. This mechanism effectively ensures the robot's operating stability in a complex dynamic water flow environment, avoids loss of control due to sudden changes in water flow or environmental changes, and improves the reliability and safety of task execution.

[0114] Example 4

[0115] Please refer to Figure 1 , specifically:

[0116] In an embodiment, the water flow change prediction module is used to construct a water flow change prediction model based on the long short-term memory network LSTM algorithm, and the water flow change prediction model is used to predict and obtain water flow data at a future time point t+k, wherein the water flow data includes water flow velocity , Water flow direction and turbulence intensity ;

[0117] According to the underwater robot operation database, historical environmental data and historical underwater robot data during the underwater robot operation process are obtained to construct a multidimensional data vector H. The multidimensional data vector H is specifically expressed in the form of:

[0118] ;

[0119] The multidimensional data vector H is input into the water flow change prediction model for model training, and the mean square error method MSE is used to optimize the parameters of the water flow change prediction model;

[0120] Using the trained water flow change prediction model, the environmental data collected in real time during the operation of the underwater robot is input into the water flow change prediction model to obtain the water flow data at the future time point t+k, and construct a water flow data time series set G, which includes the water flow velocity feature vector , Water flow direction and turbulence intensity .

[0121] In the embodiment, through the water flow change prediction model constructed based on the long short-term memory network LSTM algorithm, the system can efficiently process and predict the change of water flow at future time points, especially in complex and dynamic underwater environments. The change of water flow is usually affected by multiple factors, such as water flow speed, direction and turbulence intensity, etc. These factors are crucial to the stability and operation tasks of underwater robots. By collecting historical environmental data and historical robot data, the system can construct a multidimensional data vector H and input it into the water flow change prediction model for training, optimize the parameters of the water flow change prediction model, and the trained model can effectively predict the water flow data at the future time point t+k, including the water flow speed feature vector , Water flow direction and turbulence intensity The core advantage of this water flow change prediction model is to understand the future water flow change trend in advance, so as to provide accurate data support for subsequent thrust adjustment, attitude control and mission planning. When the system predicts that the water flow will change suddenly or in an extreme situation in the future, it can respond in advance and dynamically adjust the robot's behavior to ensure its stable operation. By constructing the water flow data time series set G, the system can obtain and update future water flow data in real time, providing a reliable basis for the intelligent control of the robot, thereby improving the underwater robot's ability to cope with changes in complex water flow environments, reducing the risk of loss of control due to sudden changes in water flow, and enhancing the robot's operating efficiency and safety in unknown or dynamic environments.

[0122] Example 5

[0123] Please refer to Figure 1 ,Specifically: the thrust optimization module includes a thrust calculation unit, an objective function construction unit and a thrust allocation unit;

[0124] The thrust calculation unit is used to obtain the target thrust of the underwater robot using the robot dynamics model based on the operation data set S and the water flow data time series set G, in combination with the preset underwater robot operation task. and compensating thrust ,Among these, the underwater robot operation task refers to reach the target location and maintain the current operation speed;

[0125] The target thrust The method of obtaining is:

[0126] ;

[0127] In the formula, represents the mass matrix of the underwater robot, represents the target acceleration eigenvector of the underwater robot, represents the control matrix, represents the velocity vector of the underwater robot, represents the resistance matrix, Represents the current position vector of the underwater robot;

[0128] The mass matrix M of the underwater robot represents the mass distribution of each part of the underwater robot, and is obtained from the underwater robot manufacturer database;

[0129] The target acceleration feature vector of the underwater robot Obtained through underwater robot mission planning database;

[0130] The control matrix It represents the resistance characteristics between the underwater robot and the water flow, reflecting the impedance of the underwater robot to the water flow at different speeds, obtained through simulation experiments;

[0131] The resistance matrix R represents the resistance of the underwater robot when it moves in water, and is obtained from the database in the robot design stage;

[0132] The compensating thrust The method of obtaining is:

[0133] ;

[0134] In the formula, represents the water flow influence coefficient, Indicates a future time point The water velocity vector, represents the velocity vector of the underwater robot;

[0135] The water flow influence coefficient Obtained through the use of fluid dynamics modeling;

[0136] According to the target thrust and compensating thrust Add together to get the underwater robot's adjusted thrust .

[0137] The objective function construction unit is used to construct an objective function J(f) for the purpose of minimizing energy consumption and to set constraints. The objective function J(f) is specifically expressed as:

[0138] ;

[0139] in, represents the total number of thrusters of the underwater robot, represents the propulsion efficiency matrix of the underwater robot, The underwater robot The thrust vector of each thruster, The underwater robot The transpose of the thrust vector of each thruster, The underwater robot Energy consumption of each thruster, q={1, 2, 3, ..., m};

[0140] The propulsion efficiency matrix of the underwater robot Obtained through the underwater robot manufacturer database;

[0141] The constraint condition refers to the thrust constraint, that is, the thrust vector of each thruster ≤Maximum thrust vector .

[0142] The thrust distribution unit is used to obtain an optimal thrust distribution scheme according to a genetic algorithm. The specific process of obtaining the optimal thrust distribution scheme is as follows:

[0143] Adjust the thrust according to the underwater robot , randomly generate g thrust allocation schemes, and for each thrust allocation scheme, obtain the objective function value of each thrust allocation scheme ;

[0144] And according to the objective function value of each thrust distribution scheme , calculate the probability of selecting the thrust allocation scheme , where the probability of selecting the thrust allocation scheme is The calculation method is:

[0145] ;

[0146] In the formula, represents the probability of selecting the jth thrust allocation scheme, represents the objective function value of the j-th thrust distribution scheme, represents the total objective function value of all thrust distribution schemes, ;

[0147] Based on the probability of selecting the thrust distribution scheme obtained , using the roulette wheel selection method, the probability of selecting all thrust allocation schemes Map to the range {0, 1} and generate a random number r, where r∈{0, 1}, and accumulate the probability of selection of each thrust allocation scheme , until the probability of selecting the thrust allocation scheme is Greater than or equal to the random number r, output the probability of selection The thrust allocation scheme is greater than or equal to the random number r, and a crossover operation is performed, wherein the crossover operation refers to the thrust allocation strategy of randomly exchanging two thrust allocation schemes using a single-point exchange method, including the thrust sizes of different thrusters;

[0148] Calculate the objective function value of the thrust distribution scheme after crossover operation , and the objective function value of the thrust distribution scheme after the crossover operation The objective function value of the thrust distribution scheme before the crossover operation For comparison, if the objective function value of the thrust distribution scheme after the cross operation is Greater than the objective function value of the thrust distribution scheme before the crossover operation , the thrust allocation scheme after the crossover operation is retained to obtain multiple thrust allocation schemes, and the thrust allocation scheme is iteratively updated repeatedly until the objective function value of the thrust allocation scheme no longer changes. At this time, the optimal thrust allocation scheme is output.

[0149] In the embodiment, through the thrust optimization module, the underwater robot can achieve efficient operation tasks in a dynamic underwater environment, especially maintain stability and reduce energy consumption in a complex fluid environment. First, through the target thrust calculation unit, the system can calculate the target thrust according to the mass matrix of the underwater robot. , target acceleration characteristic vector , control matrix C, velocity vector The thrust calculation unit can be used to calculate the target thrust required by the robot for specific tasks, such as reaching the target location and maintaining the operating speed, by using the resistance matrix R. This ensures that the robot can adapt to different operating requirements and environmental conditions when performing tasks, optimize thrust to reduce unnecessary energy consumption, and the compensating thrust calculation unit further improves the adaptability of the system. When the robot is affected by external factors such as water flow changes and turbulence, the water flow influence coefficient And combined with future time points The water velocity vector and the robot velocity vector , the system is able to calculate the compensating thrust , thereby adjusting the thrust distribution of the robot to ensure that the robot maintains stable operation in various complex environments and avoids energy waste caused by external interference. In addition, the objective function construction unit sets thrust constraints and minimizes energy consumption goals, so that the system can optimize thrust distribution and reduce energy consumption while ensuring mission success. The thrust optimization process further realizes intelligent optimization of thrust distribution through genetic algorithms. The genetic algorithm randomly generates multiple thrust distribution schemes and performs fitness evaluation, and uses roulette selection and crossover operations to continuously optimize the thrust distribution scheme until the optimal thrust distribution scheme is found. Such an intelligent optimization process not only improves the operating efficiency, but also reduces the energy consumption of the robot. In summary, the thrust optimization module ensures that the underwater robot can efficiently complete complex operating tasks, minimize energy consumption, extend the operating time, and significantly improve the stability and reliability of the robot in a dynamic environment through intelligent calculation and adaptive adjustment. This optimization method has extremely high value in practical applications, especially in high-demand tasks such as deep-sea exploration and resource development, and can provide continuous and stable support.

[0150] Example 6

[0151] Please refer to Figure 1 Specifically: the intelligent control module is used to transmit the optimal thrust distribution plan to the underwater robot control system, generate a thrust distribution instruction, and transmit the thrust distribution instruction to the underwater robot to control the underwater robot to perform underwater operations.

[0152] In the embodiment, through the intelligent control module, the system can accurately generate thrust distribution instructions according to the optimal thrust distribution plan, and transmit them to the underwater robot control system in real time. This method can ensure that the underwater robot dynamically adjusts the thrust distribution according to task requirements and environmental changes. The precise control of the thrust distribution instructions improves the stability and operation efficiency of the robot in complex underwater environments, especially when facing water flow changes, turbulence and reverse water flow, and can avoid operational errors and energy waste caused by insufficient or excessive thrust of the robot.

[0153] Example 7

[0154] Please refer to Figure 2 ,Specifically: A method for adaptive control of underwater robots based on data analysis, comprising the following steps,

[0155] Step 1: deploy an intelligent sensor group on the underwater robot to collect environmental data and underwater robot data during the operation of the underwater robot in real time, and construct an operation data set S;

[0156] Step 2: Obtain the acceleration fluctuation coefficient based on the operation data set S and attitude angular rate fluctuation coefficient , and perform summary calculation to obtain the stability score WD of the underwater robot, and preset the stability threshold WDYZ, compare and analyze the stability score WD and the stability threshold WDYZ, and evaluate the operating status of the underwater robot. If the stability score WD is greater than or equal to the stability threshold WDYZ, the operating status of the underwater robot is determined to be abnormal, and the water flow change prediction module is started;

[0157] Step 3: Collect historical environmental data and historical underwater robot data during the underwater robot operation process, and use the long short-term memory network LSTM algorithm to build a water flow change prediction model. Combined with the historical environmental data, obtain the water flow data at the future time point t+k, and build a water flow data time series set G;

[0158] Step 4: Based on the operation data set S and the water flow data time series set G, combined with the underwater robot operation task, use the robot dynamics model to obtain the target thrust of the underwater robot and compensating thrust , and according to the target thrust of the underwater robot and compensating thrust , obtain the underwater robot to adjust the thrust , and use genetic algorithm to obtain the optimal thrust distribution scheme of the underwater robot;

[0159] Step 5: Generate control instructions based on the optimal thrust distribution plan to control the underwater robot to distribute thrust and perform the operation task.

[0160] In the embodiment, the operation stability and efficiency of the underwater robot are improved by integrating the intelligent sensor group and the deep learning algorithm. First, the environmental data and robot data of the underwater robot during operation are collected in real time to establish an operation data set S to provide comprehensive basic data for subsequent analysis. and attitude angular rate fluctuation coefficient , accurately judge the robot's operating status, discover abnormalities in time and start the water flow change prediction module to ensure the safe operation of the robot in a complex environment, and in the thrust optimization module, obtain the optimal thrust distribution plan through the combination of the robot dynamics model and the genetic algorithm, which further improves the robot's operating efficiency. Finally, generate control instructions to ensure that the robot can perform tasks according to the optimal thrust distribution plan and achieve the goals of energy saving, high efficiency and stable operation.

[0161] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive control system for an underwater robot based on data analysis, characterized in that: It includes data acquisition module, stability assessment module, water flow change prediction module, thrust optimization module and intelligent control module; The data acquisition module is used to deploy an intelligent sensor group on the underwater robot to collect environmental data and underwater robot data during the operation of the underwater robot in real time and construct an operation data set S; The stability evaluation module is used to obtain the acceleration fluctuation coefficient based on the operation data set S. and attitude angular rate fluctuation coefficient , and perform summary calculation to obtain the stability score WD of the underwater robot, and preset the stability threshold WDYZ, compare and analyze the stability score WD and the stability threshold WDYZ, and evaluate the operating status of the underwater robot. If the stability score WD is greater than or equal to the stability threshold WDYZ, the operating status of the underwater robot is determined to be abnormal, and the water flow change prediction module is started; Acceleration Fluctuation Coefficient The method of obtaining is: ； In the formula, , and Respectively indicate time points The acceleration of the underwater robot on the x-axis, y-axis and z-axis is , and They represent the mean acceleration of the underwater robot on the x-axis, y-axis, and z-axis, respectively, n represents the total time point of data collection, i={1, 2, 3, ..., n}; Attitude angular rate fluctuation coefficient The method of obtaining is: ； In the formula, , and Respectively represent the pitch angle , Roll Angle and yaw angle The weight coefficient of , and Respectively represent the pitch angle , Roll Angle and yaw angle The standard deviation of the angular velocity; The stability score WD is obtained as follows: ； In the formula, and Denote acceleration fluctuation coefficients and the weight coefficient of the attitude angular rate fluctuation coefficient ZT, represents the first correction constant; The water flow change prediction module is used to collect historical environmental data and historical underwater robot data during the operation of the underwater robot, and use the long short-term memory network LSTM algorithm to build a water flow change prediction model. Combined with the historical environmental data, it obtains the water flow data at the future time point t+k and constructs the water flow data time series set G; The thrust optimization module is used to obtain the target thrust of the underwater robot based on the operation data set S and the water flow data time series set G, combined with the underwater robot operation task, and using the robot dynamics model. and compensating thrust , and according to the target thrust of the underwater robot and compensating thrust , obtain the underwater robot to adjust the thrust , and use genetic algorithm to obtain the optimal thrust distribution scheme of the underwater robot; The intelligent control module is used to generate control instructions based on the optimal thrust distribution plan, control the underwater robot to distribute thrust, and perform operating tasks.

2. The underwater robot adaptive control system based on data analysis according to claim 1, characterized in that: The data acquisition module is used to deploy an intelligent sensor group on the underwater robot to sense the water flow changes in the underwater robot's operating environment and collect environmental data and underwater robot data during the underwater robot's operation in real time. The intelligent sensor group includes a water flow sensor, a turbulence detector, a high-precision GPS, and an inertial measurement unit (IMU). Environmental data including water velocity vector , Water flow direction and turbulence intensity , where the water velocity vector The specific manifestations are: ； in, , and Respectively represent the flow rate of water in the x-axis, y-axis and z-axis; The underwater robot data includes the acceleration feature vector of the underwater robot , underwater robot velocity vector , position vector , Pitch angle , Roll Angle and yaw angle , where the acceleration eigenvector and the underwater robot velocity vector The specific manifestations are: ； ； in, , and Respectively represent the acceleration of the underwater robot on the x-axis, y-axis and z-axis, Respectively represent the movement speed of the underwater robot on the x-axis, y-axis and z-axis; Based on the acquired environmental data and underwater robot data, an operation data set S is constructed.

3. The underwater robot adaptive control system based on data analysis according to claim 2, characterized in that: The water flow change prediction module is used to build a water flow change prediction model based on the long short-term memory network LSTM algorithm, and use the water flow change prediction model to predict and obtain the water flow data at the future time point t+k, where the water flow data includes the water flow velocity , Water flow direction and turbulence intensity ; According to the underwater robot operation database, the historical environment data and historical underwater robot data during the underwater robot operation process are obtained to construct a multidimensional data vector H. The specific form of the multidimensional data vector H is: ； The multidimensional data vector H is input into the water flow change prediction model for model training, and the mean square error method MSE is used to optimize the parameters of the water flow change prediction model; Using the trained water flow change prediction model, the environmental data collected in real time during the operation of the underwater robot is input into the water flow change prediction model to obtain the water flow data at the future time point t+k, and construct the water flow data time series set G, which includes the water flow velocity feature vector , Water flow direction and turbulence intensity .

4. The underwater robot adaptive control system based on data analysis according to claim 3, characterized in that: The thrust optimization module includes a thrust calculation unit, an objective function construction unit, and a thrust allocation unit; The thrust calculation unit is used to obtain the target thrust of the underwater robot using the robot dynamics model based on the operation data set S and the water flow data time series set G, combined with the preset underwater robot operation task. and compensating thrust ,Among these, the underwater robot operation task refers to reach the target location and maintain the current operation speed; Target thrust The method of obtaining is: ； In the formula, represents the mass matrix of the underwater robot, represents the target acceleration eigenvector of the underwater robot, represents the control matrix, represents the velocity vector of the underwater robot, represents the resistance matrix, Represents the current position vector of the underwater robot; Compensation thrust The method of obtaining is: ； In the formula, represents the water flow influence coefficient, Indicates a future time point The water velocity vector, represents the velocity vector of the underwater robot; Push the target and compensating thrust Add together to get the underwater robot's adjusted thrust .

5. The underwater robot adaptive control system based on data analysis according to claim 4, characterized in that: The objective function construction unit is used to construct the objective function J(f) and set up constraints for the purpose of minimizing energy consumption. The specific form of the objective function J(f) is: ； in, represents the total number of thrusters of the underwater robot, represents the propulsion efficiency matrix of the underwater robot, The underwater robot The thrust vector of each thruster, The underwater robot The transpose of the thrust vector of each thruster, The underwater robot Energy consumption of each thruster, q={1, 2, 3, ..., m}; Propulsion efficiency matrix for underwater robots Obtained through the underwater robot manufacturer database; The constraints refer to the thrust constraints, that is, the thrust vector of each thruster ≤Maximum thrust vector .

6. The underwater robot adaptive control system based on data analysis according to claim 5, characterized in that: The thrust distribution unit is used to obtain the optimal thrust distribution plan based on the genetic algorithm. The specific process of obtaining the optimal thrust distribution plan is as follows: Adjust thrust according to underwater robot , randomly generate g thrust allocation schemes, and for each thrust allocation scheme, obtain the objective function value of each thrust allocation scheme ; And according to the objective function value of each thrust distribution scheme , calculate the probability of selecting the thrust allocation scheme , where the probability of selecting the thrust allocation scheme is The calculation method is: ； In the formula, represents the probability of selecting the jth thrust allocation scheme, represents the objective function value of the j-th thrust distribution scheme, represents the total objective function value of all thrust distribution schemes, ; Based on the probability of selecting the thrust distribution scheme obtained , using the roulette wheel selection method, the probability of selecting all thrust allocation schemes Map to the range {0, 1} and generate a random number r, where r∈{0, 1}, and accumulate the probability of selection of each thrust allocation scheme , until the probability of selecting the thrust allocation scheme is Greater than or equal to the random number r, output the probability of selection The thrust allocation scheme is greater than or equal to the random number r, and a crossover operation is performed, wherein the crossover operation refers to the thrust allocation strategy of randomly exchanging two thrust allocation schemes using a single-point exchange method, including the thrust sizes of different thrusters; Calculate the objective function value of the thrust distribution scheme after the crossover operation , and the objective function value of the thrust distribution scheme after the crossover operation The objective function value of the thrust distribution scheme before the crossover operation For comparison, if the objective function value of the thrust distribution scheme after the cross operation is Greater than the objective function value of the thrust distribution scheme before the crossover operation , the thrust allocation scheme after the crossover operation is retained to obtain multiple thrust allocation schemes, and the thrust allocation scheme is iteratively updated repeatedly until the objective function value of the thrust allocation scheme no longer changes. At this time, the optimal thrust allocation scheme is output.

7. The underwater robot adaptive control system based on data analysis according to claim 6, characterized in that: The intelligent control module is used to transmit the optimal thrust distribution plan to the underwater robot control system, generate thrust distribution instructions, and transmit the thrust distribution instructions to the underwater robot to control the underwater robot to perform underwater operations.

8. An underwater robot adaptive control method based on data analysis, comprising the following steps: Step 1: deploy an intelligent sensor group on the underwater robot to collect environmental data and underwater robot data during the operation of the underwater robot in real time, and construct an operation data set S; Step 2: Obtain the acceleration fluctuation coefficient based on the operation data set S and attitude angular rate fluctuation coefficient , and perform summary calculation to obtain the stability score WD of the underwater robot, and preset the stability threshold WDYZ, compare and analyze the stability score WD and the stability threshold WDYZ, and evaluate the operating status of the underwater robot. If the stability score WD is greater than or equal to the stability threshold WDYZ, the operating status of the underwater robot is determined to be abnormal, and the water flow change prediction module is started; Acceleration Fluctuation Coefficient The method of obtaining is: ； In the formula, , and Respectively indicate time points The acceleration of the underwater robot on the x-axis, y-axis and z-axis is , and They represent the mean acceleration of the underwater robot on the x-axis, y-axis, and z-axis, respectively, n represents the total time point of data collection, i={1, 2, 3, ..., n}; Attitude angular rate fluctuation coefficient The method of obtaining is: ； In the formula, , and Respectively represent the pitch angle , Roll Angle and yaw angle The weight coefficient of , and Respectively represent the pitch angle , Roll Angle and yaw angle The standard deviation of the angular velocity; The stability score WD is obtained as follows: ； In the formula, and Denote acceleration fluctuation coefficients and the weight coefficient of the attitude angular rate fluctuation coefficient ZT, represents the first correction constant; Step 3: Collect historical environmental data and historical underwater robot data during the underwater robot operation process, and use the long short-term memory network LSTM algorithm to build a water flow change prediction model. Combined with the historical environmental data, obtain the water flow data at the future time point t+k, and build a water flow data time series set G; Step 4: Based on the operation data set S and the water flow data time series set G, combined with the underwater robot operation task, use the robot dynamics model to obtain the target thrust of the underwater robot and compensating thrust , and according to the target thrust of the underwater robot and compensating thrust , obtain the underwater robot to adjust the thrust , and use genetic algorithm to obtain the optimal thrust distribution scheme of the underwater robot; Step 5: Generate control instructions based on the optimal thrust distribution plan to control the underwater robot to distribute thrust and perform the task.

Images