Propulsion control method, system and device applied to underwater robot

Abstract

The application relates to the technical field of posture control, in particular to a propelling control method, system and device applied to an underwater robot. The method obtains sigma points of an unscented Kalman filtering process through an environmental system state vector of the robot, and obtains an adjustment limiting vector of a sampling moment according to numerical fluctuation of different dimensions; a confusion degree evaluation index is obtained according to unstable conditions of continuous changes of the adjustment limiting vector; adaptive regulation and control of switching gain is carried out through deviation conditions of a current and initial limiting range and the confusion degree evaluation, and the confusion degree evaluation is combined with adjustment of a sampling frequency to re-output a control signal. In the adjustment process of the switching gain and the sampling frequency through continuous environmental interference confusion, the application improves the gain to make the system quickly converge while considering limiting adjustment, reduces the jitter condition of the control output, and makes the adjustment control of the posture of each propelling action of the AUV more accurate and efficient.

Description

Technical Field

[0001] This invention relates to the field of attitude control technology, and more specifically to propulsion control methods, systems and devices for underwater robots. Background Technology

[0002] Autonomous underwater vehicles (AUVs) are highly autonomous, but they are also susceptible to hazards such as entanglement with marine life, disturbance from underwater currents, and seawater corrosion during underwater operations. Therefore, during autonomous operation, AUVs need to be aware of the surrounding underwater environment and adjust their navigation attitude accordingly to ensure that they can actively avoid dangers.

[0003] Existing technologies are mainly divided into two types: one based on data-driven and the other based on model analysis. As integrated circuits are being developed to a higher scale, the high computing power requirements of model analysis are no longer a bottleneck restricting its deployment. Therefore, the accurate analysis results brought by model analysis can greatly improve the safety of AUVs in underwater operations.

[0004] In existing technologies for AUV underwater operations, the unpredictable and rapid changes in the underwater environment cause lag issues in AUV attitude control. Unscented Kalman filtering is commonly used to detect the real-time attitude of the AUV and the real-time impact of the underwater environment on its attitude. The switching gain of sliding mode control is adjusted in real-time based on the predicted state variables to reduce output jitter caused by excessive control adjustment frequency. However, unscented Kalman filtering generates multiple low-probability results during the prediction process, making it difficult to further optimize subsequent calculation delays. The large errors in the predicted output range also affect the accuracy of attitude control adjustments. Summary of the Invention

[0005] To address the technical problems of existing unscented Kalman filtering technologies, which generate multiple low-probability results during the prediction process, making further optimization of subsequent calculation delays difficult, and the large fluctuations in the prediction output range affecting the accuracy of control attitude adjustment, the present invention aims to provide a propulsion control method, system, and device for underwater robots. The specific technical solution adopted is as follows:

[0006] This invention provides a propulsion control method for underwater robots, the method comprising:

[0007] Obtain the environmental system state vector of the underwater robot at each sampling time from the initial time to the current time; perform unscented Kalman filtering on the environmental system state vector at each sampling time to obtain the sigma point at the sigma point generation stage at each sampling time, as well as the predicted state vector at the current time.

[0008] Based on the fluctuation of the numerical distribution in each dimension of the sigma point at each sampling time, the adjustment constraint vector at each sampling time is obtained;

[0009] Based on the distribution deviation between the changing trend of the constraint vector in all sampling times before the current time and the overall trend of the constraint vector adjustment from the initial time to the current time, the disorder evaluation index at the current time is obtained.

[0010] Based on the current environmental system state vector and the predicted state vector, the switching gain at the current moment is obtained through the sliding mode controller; based on the similarity between the adjustment constraint vector at the current moment and the initial moment, combined with the disorder evaluation index and the switching gain at the current moment, the adjustment switching gain at the current moment is obtained.

[0011] The sampling frequency is adjusted based on the disorder evaluation index at the current moment to obtain the adjusted sampling frequency at the current sampling moment; the adjusted sampling frequency and the adjusted switching gain at the current moment are then re-input into the sliding mode controller to obtain the control signal for the next moment.

[0012] Furthermore, the method for obtaining the adjustment constraint vector includes:

[0013] For any sampling time, the constraint value for each dimension is obtained based on the degree of concentration of the values ​​in each dimension of the sigma point;

[0014] Arrange the constraint values ​​of each dimension at the sampling time according to the distribution order of the dimensions in the environmental system state vector to obtain the adjustment constraint vector at that sampling time.

[0015] Furthermore, the method for obtaining the limit value includes:

[0016] For any dimension at the sampling time, calculate the mean of the values ​​of all elements in that dimension at the sigma point at that sampling time, and use it as the distribution mean of that dimension; calculate the standard deviation of the values ​​of all elements in that dimension at the sigma point at that sampling time, and use it as the distribution deviation of that dimension.

[0017] Subtract the mean of the distribution from the value of each element in the dimension at the sigma point at the sampling time, and then divide by the distribution deviation value to obtain the standardized value of each element in the dimension.

[0018] After calculating the fourth power of the standardized value of each element in this dimension, the cumulative value is calculated to obtain the constraint value of this dimension.

[0019] Furthermore, the method for obtaining the disorder evaluation index includes:

[0020] Based on the distribution of the adjustment constraint vector at the previous time-series upsampling time, the judgment curve at the current time is obtained; based on the distribution trend of the adjustment constraint vector between the initial time and the current time, the error advancement line at the current time is obtained.

[0021] The shortest distance between the data point corresponding to each sampling moment in the judgment curve and the error advancement line is taken as the deviation distance value at each sampling moment on the judgment curve.

[0022] Integrate the deviation distance value at each sampling time on the judgment curve from the initial time to the current time to obtain the limit range value at the current time;

[0023] Multiply the time difference between the initial time and the current time by the mean of all deviation distance values ​​in the judgment curve to obtain the error change value at the current time.

[0024] The difference between the current limit range value and the error change value is used as the disorder evaluation index at the current moment.

[0025] Furthermore, the method for obtaining the judgment curve includes:

[0026] Each element in the adjustment constraint vector is used as a dimension to establish a multidimensional spatial coordinate system. The adjustment constraint vector at each sampling time is then mapped to the multidimensional spatial coordinate system to obtain the data points at each sampling time in the multidimensional spatial coordinate system.

[0027] Connect the data points in chronological order of sampling time to obtain the judgment curve.

[0028] Furthermore, the method for obtaining the error propagation line includes:

[0029] The line connecting the data points corresponding to the initial time and the current time in the multidimensional coordinate system is used as the error propagation line.

[0030] Furthermore, the method for obtaining the adjusted switching gain includes:

[0031] The sum of the cosine similarity of the adjustment constraint vector between the initial time and the current time and the preset adjustment value is used as the stability adjustment coefficient at the current time.

[0032] The disorder evaluation index at the current moment is normalized and range-mapped to obtain the disorder adjustment coefficient at the current moment.

[0033] The product of the switching gain and the current chaos adjustment coefficient is used as the current gain adjustment value; the sum of the current switching gain and the gain adjustment value is used as the current initial adjustment gain.

[0034] The product of the initial adjustment gain and the stable adjustment coefficient at the current moment is used as the adjustment switching gain at the current moment.

[0035] Furthermore, the method for obtaining the adjusted sampling frequency includes:

[0036] The product of the current sampling frequency and the disorder adjustment coefficient is used as the frequency adjustment value at the current time; the sum of the current sampling frequency and the frequency adjustment value is used as the initial adjustment frequency at the current time.

[0037] When the initial adjustment frequency is greater than or equal to the preset lower frequency limit and less than or equal to the preset upper frequency limit, the initial adjustment frequency is used as the adjustment sampling frequency at the current sampling time.

[0038] When the initial adjustment frequency is less than the preset lower limit, the preset lower limit will be used as the adjustment sampling frequency at the current sampling time.

[0039] When the initial adjustment frequency is greater than the preset frequency upper limit, the preset frequency upper limit will be used as the adjustment sampling frequency at the current sampling time.

[0040] The present invention also provides a propulsion control system for an underwater robot, the system comprising:

[0041] The data acquisition module is used to acquire the environmental system state vector of the underwater robot at each sampling time from the initial time to the current time; and to perform unscented Kalman filtering on the environmental system state vector at each sampling time to obtain the sigma point at the sigma point generation stage at each sampling time, as well as the predicted state vector at the current time.

[0042] The adjustment constraint analysis module is used to obtain the adjustment constraint vector for each sampling time based on the fluctuation of the numerical distribution in each dimension of the sigma point at each sampling time.

[0043] The trend disorder analysis module is used to obtain the disorder evaluation index at the current moment by comparing the distribution deviation between the changing trend of the continuously adjusted constraint vector at all sampling moments before the current moment and the overall trend of the adjusted constraint vector from the initial moment to the current moment.

[0044] The switching gain adjustment module is used to obtain the switching gain at the current moment through the sliding mode controller based on the current environmental system state vector and the predicted state vector; and to obtain the adjusted switching gain at the current moment based on the similarity between the adjustment constraint vector at the current moment and the initial moment, combined with the disorder evaluation index at the current moment and the switching gain.

[0045] The control signal output module is used to adjust the sampling frequency at the current moment according to the disorder evaluation index at the current moment, and obtain the adjusted sampling frequency at the current sampling moment; based on the adjusted sampling frequency and the adjusted switching gain at the current moment, it is re-input into the sliding mode controller and outputs the control signal for the next moment.

[0046] The present invention also provides a propulsion control device for an underwater robot, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-described propulsion control methods for an underwater robot.

[0047] The present invention has the following beneficial effects:

[0048] This invention considers the short-term passive attitude stabilization characteristic of AUVs caused by impacts from underwater environmental disturbances. During the sigma point generation stage of unscented Kalman filtering of the environmental system state vector, the adjustment constraint vector at the sampling time is obtained by measuring the numerical fluctuations of different environmental types in the sigma points. This measures the margin of the AUV's current attributes in the propulsion attitude control process, limiting subsequent adjustments to prevent AUV attitude loss of control. Furthermore, based on the instability of the continuously changing adjustment constraint vector, a disorder evaluation index is obtained to characterize the degree of instability of the predicted system at the current moment. This dynamic assessment of the system state provides a basis for subsequent dynamic adjustment strategies. Adaptive control of the switching gain is performed based on the deviation between the current and initial constraint ranges and the disorder evaluation. Combined with the disorder evaluation, the sampling frequency is adjusted. When increasing the gain to achieve rapid system convergence, constraint adjustments are made to reduce control output jitter, making the attitude adjustment control of the AUV more precise and efficient for each propulsion maneuver. Attached Figure Description

[0049] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating a propulsion control method for an underwater robot, provided as an embodiment of the present invention;

[0051] Figure 2 This is a schematic diagram of the sigma point matrix distribution corresponding to a sampling time, provided in one embodiment of the present invention;

[0052] Figure 3A schematic diagram of a two-dimensional judgment curve and error progression line provided in one embodiment of the present invention;

[0053] Figure 4 This is a schematic diagram of the integration between a two-dimensional judgment curve and an error propagation line, provided as an embodiment of the present invention.

[0054] Figure 5 This is a schematic diagram of an output control line under switching gain provided in one embodiment of the present invention;

[0055] Figure 6 This is a schematic diagram of an adjusted output control line provided in one embodiment of the present invention. Detailed Implementation

[0056] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a propulsion control method, system, and apparatus for underwater robots proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0058] The following description, in conjunction with the accompanying drawings, details a specific solution for a propulsion control method, system, and device for underwater robots provided by the present invention.

[0059] The Unscented Kalman Filter (UKF) is an improved Kalman filtering method primarily used for state estimation in nonlinear systems. Compared to the traditional Kalman Filter (KF), the UKF uses an unscented transform to more accurately handle the prediction and update process of nonlinear systems. Its main processes include: Initialization phase: setting the initial state estimate and covariance matrix. Sigma point generation phase: generating a set of sigma points from the current state estimate and covariance matrix. State prediction phase: predicting the state at the next time step for each sigma point using the system's nonlinear dynamic model. The predicted state and covariance are then updated using the weighted average of these sigma points. Measurement prediction phase: mapping the predicted state's sigma points to the observation space using a nonlinear measurement model to obtain the predicted observations. Kalman gain calculation phase: calculating the Kalman gain based on the predicted observations and actual measurements, used to update the state estimate. State estimate update phase: updating the state estimate using the Kalman gain and adjusting the covariance matrix. Iterative process: Repeat the prediction and update steps to continuously refine the state estimate.

[0060] Sliding Mode Control (SMC) is a robust control method widely used to handle dynamic systems with uncertainties and external disturbances. A sliding mode controller designs a "sliding surface" on which the system state can slide. The process mainly includes sliding surface design, control law calculation, and gain adjustment to achieve system stability and desired dynamic performance. The input to a sliding mode controller is typically the system's state variables; based on these state variables, the controller calculates the control input. The output is usually the calculated control signal used to drive the input of the controlled system.

[0061] Example 1:

[0062] Please see Figure 1 The diagram illustrates a propulsion control method for an underwater robot according to an embodiment of the present invention, which includes the following steps:

[0063] S1: Obtain the environmental system state vector of the underwater robot at each sampling time from the initial time to the current time; perform unscented Kalman filtering on the environmental system state vector at each sampling time to obtain the sigma point at the sigma point generation stage at each sampling time, as well as the predicted state vector at the current time.

[0064] Based on the tasks that the AUV needs to perform underwater, the AUV is selected and assembled, then deployed to a predetermined water area for underwater operations. Simultaneously, the sensing unit is activated to monitor data and obtain real-time system status in the underwater environment. In this embodiment of the invention, all sensors are first time-series aligned to ensure that all sampled data are acquired at each sampling moment. The sampling frequency is set to a range of 0.5Hz-10Hz, allowing for different sampling frequencies to be used in various complex underwater environments.

[0065] In this embodiment of the invention, the data types of the collected parameters include: water pressure (P), unit MPa, megapascal; depth (D), unit m, meter; speed K, unit m / s, meter per second; forward / backward (J): unit m / s, meter per second, corresponding to the linear motion of the AUV along the forward / backward direction, degree of freedom: translation along the X-axis; translation (S): unit m / s, meter per second, corresponding to the lateral motion of the AUV along the left / right direction, degree of freedom: translation along the Y-axis; snorkeling (H): unit m / s, meter per second, corresponding to the vertical motion of the AUV along the up / down direction, degree of freedom: translation along the Z-axis; pitch (Q): unit m / s², meter per square second, corresponding to the rotation of the AUV about its lateral axis, affecting the lifting or sinking of the front and rear ends of the AUV, degree of freedom: rotation along the X-axis; roll (R): unit m / s², meter per square second, corresponding to the rotation of the AUV about its lateral axis, affecting the lifting or sinking of the front and rear ends of the AUV, degree of freedom: rotation along the X-axis; yaw (R): unit m / s², meter per square second, corresponding to the rotation of the AUV about its lateral axis, affecting the lifting or sinking of the front and rear ends of the AUV, degree of freedom: rotation along the X-axis; roll (R): unit m / s², meter per square second, corresponding to the rotation of the AUV about its lateral axis. Meters per square second corresponds to the rotation of the AUV about its longitudinal axis, affecting the AUV's roll. Degree of freedom: rotation along the Y-axis; yaw (Y): unit. Meters per square second corresponds to the AUV's rotation around its vertical axis, affecting its heading. The degree of freedom is rotation along the Z-axis. For each sampling time, the collected data is used to construct a vector representation reflecting the current state of the environmental system, facilitating subsequent analysis. For example, the environmental system state vector for sampling time t is... The subscript T represents the vector transpose. It should be noted that the types of parameters collected can be adjusted according to the different needs of underwater robots, and no restrictions are imposed here.

[0066] At this point, the environmental system state vector is complete, which reflects the attitude parameters of the AUV. Therefore, the environmental system state vector is further input into the unscented transform module in the UKF filter. The unscented transform module can estimate a set of possible impact results based on the current environmental system state vector and its numerical error, thus obtaining the possible impact at the current moment.

[0067] In this embodiment of the invention, during the sigma point generation stage of the unscented Kalman filter, the sigma points corresponding to each sampling time are obtained, totaling... One, of which This represents the number of types of parameters in the environmental system's state vector. Please refer to [link / reference]. Figure 2It shows a schematic diagram of the sigma point matrix distribution corresponding to a sampling time according to an embodiment of the present invention, wherein the horizontal axis represents the number of sigma points, and each column corresponds to one sigma point, that is, a total of The columns, with the vertical axis representing the number of parameter types, and each row corresponding to one parameter type in the environmental system state vector, are also known as the total number of parameters. OK.

[0068] For the current moment, the sigma point of the current moment can be input into the UKF filter to obtain the predicted state vector of the next moment, that is, to obtain the predicted state of each parameter at the next moment.

[0069] S2: Based on the fluctuation of the numerical distribution in each dimension of the sigma point at each sampling time, obtain the adjustment constraint vector for each sampling time.

[0070] Analyzing the impact of the state at each moment, the varying degrees of deviation at the sigma points reflect the potential errors caused by the underwater environment at the sampling time. By measuring the range of influence of data in different dimensions, the control margin of the AUV's propulsion attitude in each dimension is used as a constraint.

[0071] Preferably, in this embodiment of the invention, the method for obtaining the adjustment constraint vector includes:

[0072] For any sampling time, the constraint value of each dimension is obtained based on the degree of concentration of the values ​​in each dimension of the sigma point. The more concentrated the data distribution in the dimension, that is, the higher the degree of concentration of the values ​​in each parameter, the more restrictions need to be imposed when adjusting the corresponding parameter, and the smaller the space for adjusting the values ​​of the corresponding dimension.

[0073] In this embodiment of the invention, the method for obtaining the limit value for each dimension includes:

[0074] For any dimension at the sampling time, calculate the mean of all elements in that dimension at the sigma point at that sampling time, as the distribution mean of that dimension. Calculate the standard deviation of all elements in that dimension at the sigma point at that sampling time, as the distribution bias of that dimension. Subsequently, each element can be standardized using the mean and standard deviation of the numerical distribution in a single dimension to eliminate the scaling effect.

[0075] The standardized value of each element in that dimension is obtained by subtracting the mean from the value of each element at the sigma point at the sampling time, and then dividing by the distribution deviation. The standardized value of each element in that dimension is then raised to the fourth power, and the accumulated value is calculated to obtain the constraint value for that dimension. The fourth power amplifies the sum, emphasizing the contribution of data points far from the mean. The final sum reflects the concentration of the data distribution around the mean. As an example, the expression for the constraint value is:

[0076] In the formula, Represented as the first Limitation values ​​for each dimension, Represented as the first The total number of elements in each dimension This is represented by the value of the i-th element; Represented as the first The distribution mean of each dimension Represented as the first Distribution deviation values ​​in each dimension. Represented as the first The standardized value of the i-th element in each dimension.

[0077] Finally, the constraint values ​​of each dimension at the sampling time are arranged in the order of the distribution of dimensions in the environmental system state vector to obtain the adjustment constraint vector at that sampling time. The adjustment constraint vector reflects the attitude adjustment constraint obtained by combining all parameters at the current time.

[0078] S3: Based on the distribution deviation between the changing trend of the continuously adjusted constraint vector at all sampling times before the current time and the overall trend of the adjusted constraint vector from the initial time to the current time, the disorder evaluation index at the current time is obtained.

[0079] When an AUV operates underwater, it is subject to the impact of currents or moving objects. Due to the inherent resistance of water, the actuators of the propulsion system (including thrusters in all directions of the fuselage) must first overcome the resistance of the water and then overcome the impact of the underwater environment on the fuselage in order to propel the fuselage.

[0080] When an AUV accelerates in water, it needs to move a portion of the surrounding water along with it; the mass of this water is called the added mass. This added mass increases the AUV's effective mass, making its acceleration smaller and its motion easier to maintain, thus allowing for a greater range of motion. However, this requires the thrusters to do more work to propel the AUV and the combined weight of the added water.

[0081] Because the propulsion unit's output parameters are set to resist the previous external environment when an environmental impact is transmitted to the fuselage, the fuselage will deflect along the impact direction until the impact energy overcomes the combined weight of the fuselage and its components, propelling the assembly forward until the energy is exhausted. Therefore, analyzing various external disturbances, larger impacts result in greater attitude deviations and higher demands on propulsion control, leading to more stringent adjustment requirements. Thus, analyzing the changes in the continuously adjusting constraint vector reveals the chaotic situations requiring adjustment.

[0082] Preferably, in this embodiment of the invention, the method for obtaining the disorder evaluation index includes:

[0083] First, based on the distribution of the adjustment constraint vectors from the previous time-series upsampling times, the judgment curve for the current time is obtained. Through continuous changes in the constraint range, the stability of the fuselage under impact is reflected. In this embodiment, each element of the adjustment constraint vector is treated as a dimension to establish a multi-dimensional coordinate system. The adjustment constraint vector at each sampling time is mapped to this multi-dimensional coordinate system, obtaining the data point for each sampling time. In multi-dimensional space, a vector can be represented as a point. Connecting the data points in chronological order of the sampling times yields the judgment curve, reflecting the vector changes.

[0084] Further, based on the distribution trend of the constraint vector between the initial and current times, the error propagation line at the current time is obtained. In this embodiment of the invention, the line connecting the corresponding data points at the initial and current times in the multi-dimensional coordinate system is used as the error propagation line, which reflects the overall trend in the time series. Furthermore, the degree of interference can be reflected by judging the deviation of the curve from the error propagation line. Please refer to [link to relevant documentation]. Figure 3 It shows a schematic diagram of a two-dimensional judgment curve and error advancement line provided by an embodiment of the present invention.

[0085] Then, the shortest distance from the data point corresponding to each sampling moment on the judgment curve to the error advancement line is used to obtain the deviation distance value at each sampling moment on the judgment curve, reflecting the deviation. The deviation distance value at each sampling moment on the judgment curve is integrated from the initial moment to the current moment to obtain the current limit range value. The integration reflects the magnitude of the fluctuation of the limit range of the adjustment limit vector at each moment, that is, the magnitude of the impact of external environmental forces on the body.

[0086] The time difference between the initial moment and the current moment is further multiplied by the mean of all deviation distance values ​​in the judgment curve to obtain the error change value at the current moment. The change caused by the stable shock is explained by the product of the mean distance and the duration.

[0087] Finally, the difference between the current constraint range value and the error change value is used as the disorder evaluation index for the current moment. By using the error change value, stable shock situations are suppressed in the shock analysis, reflecting the overall degree of influence from multiple shocks at the current moment. Please refer to [link / reference]. Figure 4 It shows a schematic diagram of the integration between a two-dimensional judgment curve and an error advancement line provided by an embodiment of the present invention.

[0088] S4: Based on the current environmental system state vector and the predicted state vector, obtain the current switching gain through the sliding mode controller; based on the similarity between the current and initial adjustment constraint vectors, and combined with the current disorder evaluation index and switching gain, obtain the current adjustment switching gain.

[0089] A key advantage of sliding mode controllers is their robustness to system uncertainties and external disturbances. Switching gain is crucial for achieving this robustness; by selecting a larger switching gain, the control system can remain stable and achieve the intended control objective even under uncertainties and disturbances. Switching gain also affects the speed at which the system state reaches the sliding surface; a larger gain results in faster convergence. However, high-frequency, rapid convergence leads to output jitter, meaning rapid output changes between propulsion states and unstable machine movement. Therefore, when adjusting propulsion control, it is necessary to suppress jitter as much as possible.

[0090] First, the switching gain under normal conditions is obtained. The switching gain is typically determined by the difference between the current system state and the predicted state. In this embodiment of the invention, the current environmental system state vector and the predicted state vector predicted by UKF filtering are input to a pre-calibrated sliding mode controller to obtain the directly generated switching gain. See also... Figure 5 This diagram illustrates an output control line under switching gain according to an embodiment of the present invention.

[0091] The higher the demand for rapid response and interference resistance, i.e., the higher the disorder evaluation index, the greater the degree of correction required. Furthermore, anti-jitter limits are implemented considering the stability of the fluctuation range between the current and initial moments, and a more suitable switching gain is comprehensively adjusted.

[0092] Preferably, in this embodiment of the invention, the method for obtaining the adjustment switching gain includes:

[0093] First, the sum of the cosine similarity of the adjustment constraint vector between the initial time and the current time and the preset adjustment value is used as the stability adjustment coefficient at the current time. In this embodiment of the invention, the preset adjustment value is set to 1, which means that the stability adjustment coefficient is obtained by adding 1 to the calculated cosine similarity value. The more similar the adjustment constraint vector is, the higher the temporary stability generated at the current time, and the smaller the suppression effect on the switching gain.

[0094] The disorder evaluation index at the current moment is normalized and mapped to a normalized range to obtain the disorder adjustment coefficient at the current moment. In this embodiment of the invention, the disorder evaluation index is normalized to the [-1,1] interval using the premnmx function to ensure that overflow does not occur during the adjustment of the switching gain value of the SMC. It should be noted that the normalization method and cosine similarity calculation are techniques well known to those skilled in the art and will not be elaborated here.

[0095] Furthermore, the product of the switching gain and the current chaos adjustment coefficient is used as the current gain adjustment value, and the sum of the current switching gain and the gain adjustment value is used as the current initial adjustment gain, which represents the size of the switching gain after adjustment only by chaos adjustment requirements.

[0096] Finally, the product of the initial adjustment gain and the stable adjustment coefficient at the current moment is used as the adjustment switching gain at the current moment. By suppressing system stability and reducing jitter, a more suitable adjustment switching gain may be obtained. As an example, the expression for the adjustment switching gain is:

[0097] In the formula, Representing the current time Adjusting the switching gain, Representing the current time Switching gain, Representing the current time The disorder adjustment coefficient, Representing the current time The stability adjustment coefficient, Representing the current time Gain adjustment value, Representing the current time The initial adjustment gain.

[0098] S5: Adjust the sampling frequency at the current moment according to the disorder evaluation index at the current moment to obtain the adjusted sampling frequency at the current sampling moment; re-input the adjusted sampling frequency and the adjusted switching gain at the current moment into the sliding mode controller to obtain the control signal at the next moment.

[0099] Furthermore, considering the close relationship between sampling frequency adjustment and the real-time performance of the control strategy, a higher sampling frequency generally helps the system adjust the control input in real time and improves control accuracy, but it may also increase the computational burden. If the system has high erratic behavior or drastic state changes, it may be necessary to increase the sampling frequency to respond quickly to system changes and provide more accurate control input. Conversely, if the system is relatively stable, the sampling frequency can be reduced to decrease computation and save resources.

[0100] Therefore, the sampling frequency at the current moment is also adjusted using the disorder evaluation index. In this embodiment of the invention, the method for obtaining the adjusted sampling frequency includes:

[0101] The product of the current sampling frequency and the disorder adjustment coefficient is used as the frequency adjustment value at the current moment. The degree of adjustment is analyzed by the disorder evaluation index after normalization. Furthermore, the sum of the current sampling frequency and the frequency adjustment value is used as the initial adjustment frequency at the current moment.

[0102] Since there are preset upper and lower limits for the sampling frequency, it is necessary to determine whether the adjustment can be used. When the initial adjustment frequency is greater than or equal to the preset lower limit and less than or equal to the preset upper limit, it means that the adjusted frequency is within the frequency range, and the initial adjustment frequency is used as the adjustment sampling frequency at the current sampling time.

[0103] Therefore, if the initial adjustment frequency is less than the preset lower limit, it indicates that the adjustment is too low, and the preset lower limit is used as the adjustment sampling frequency at the current sampling time. If the initial adjustment frequency is greater than the preset upper limit, it indicates that the adjustment is too high, and the preset upper limit is used as the adjustment sampling frequency at the current sampling time. In this embodiment of the invention, the preset upper limit is 10Hz and the preset lower limit is 0.5Hz. The specific values ​​can be adjusted by the implementer according to the specific implementation scenario, and are not limited here.

[0104] In dynamic control systems, switching gain and sampling frequency interact. Adjusting the switching gain allows for better handling of unstable or complex system behavior, while adjusting the sampling frequency balances the accuracy of the system response with computational efficiency. The adjusted switching gain and sampling frequency are then re-input into the sliding mode controller to update the control signal. See also... Figure 6 This illustrates a schematic diagram of an adjusted output control line provided in one embodiment of the present invention.

[0105] In summary, this invention considers the short-term passive attitude stabilization caused by the impact of underwater environmental disturbances on the AUV, resulting in the fuselage moving under impact. During the sigma point generation stage of the unscented Kalman filter on the environmental system state vector, the adjustment constraint vector at the sampling time is obtained through the numerical fluctuations of different environmental types in the sigma points. This measures the margin of the AUV's current attributes in the propulsion attitude control process, limiting subsequent adjustments to prevent AUV attitude loss of control. Furthermore, based on the instability of the continuously changing adjustment constraint vector, a disorder evaluation index is obtained to characterize the degree of instability of the predicted system at the current moment. This dynamic evaluation of the system state provides a basis for subsequent dynamic adjustment strategies. Adaptive control of the switching gain is performed based on the deviation between the current and initial constraint ranges and the disorder evaluation. Combined with the adjustment of the sampling frequency by the disorder evaluation, constraint adjustments are made when increasing the gain to achieve rapid system convergence, reducing control output jitter and making the attitude adjustment control of the AUV more precise and efficient for each propulsion maneuver.

[0106] Example 2:

[0107] The present invention also provides a propulsion control system for an underwater robot, the system comprising:

[0108] The data acquisition module is used to acquire the environmental system state vector of the underwater robot at each sampling time from the initial time to the current time; and to perform unscented Kalman filtering on the environmental system state vector at each sampling time to obtain the sigma point at the sigma point generation stage at each sampling time, as well as the predicted state vector at the current time.

[0109] The adjustment constraint analysis module is used to obtain the adjustment constraint vector for each sampling time based on the fluctuation of the numerical distribution in each dimension of the sigma point at each sampling time.

[0110] The trend disorder analysis module is used to obtain the disorder evaluation index at the current moment by comparing the distribution deviation between the changing trend of the continuously adjusted constraint vector at all sampling moments before the current moment and the overall trend of the adjusted constraint vector from the initial moment to the current moment.

[0111] The switching gain adjustment module is used to obtain the switching gain at the current moment through the sliding mode controller based on the current environmental system state vector and the predicted state vector; and to obtain the adjusted switching gain at the current moment based on the similarity between the adjustment constraint vector at the current moment and the initial moment, combined with the disorder evaluation index at the current moment and the switching gain.

[0112] The control signal output module is used to adjust the sampling frequency at the current moment according to the disorder evaluation index at the current moment, and obtain the adjusted sampling frequency at the current sampling moment; based on the adjusted sampling frequency and the adjusted switching gain at the current moment, it is re-input into the sliding mode controller and outputs the control signal for the next moment.

[0113] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional module units as needed, that is, the internal structure of the system can be divided into different functional module units to complete all or part of the functions described above. Since the specific implementation process of the propulsion control system for underwater robots in this embodiment is the same as the specific implementation process of the propulsion control method for underwater robots described above, it will not be described in detail here.

[0114] Example 3:

[0115] The present invention also provides a propulsion control device for an underwater robot, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-described propulsion control methods for an underwater robot.

[0116] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0117] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A propulsion control method for underwater robots, characterized in that, The method includes: Obtain the environmental system state vector of the underwater robot at each sampling time from the initial time to the current time; perform unscented Kalman filtering on the environmental system state vector at each sampling time to obtain the sigma point at the sigma point generation stage at each sampling time, as well as the predicted state vector at the current time. Based on the fluctuation of the numerical distribution in each dimension of the sigma point at each sampling time, the adjustment constraint vector at each sampling time is obtained; Based on the distribution deviation between the trend of the adjustment constraint vector change of all consecutive sampling times before the current time and the overall trend of the adjustment constraint vector from the initial time to the current time, the disorder evaluation index of the current time is obtained. Based on the current environmental system state vector and the predicted state vector, the switching gain at the current moment is obtained through the sliding mode controller; based on the cosine similarity between the current moment and the initial moment of the adjustment constraint vector, combined with the current moment's disorder evaluation index and the switching gain, the adjustment switching gain at the current moment is obtained. The sampling frequency at the current moment is adjusted by using the disorder evaluation index at the current moment to obtain the adjusted sampling frequency at the current sampling moment; based on the adjusted sampling frequency at the current moment, the adjusted switching gain is re-input into the sliding mode controller to obtain the control signal for the next moment. The method for obtaining the adjustment constraint vector includes: For any given sampling time, based on the degree of concentration of the values ​​in each dimension of the sigma point, the constraint value for each dimension is obtained; the constraint values ​​for each dimension at that sampling time are arranged in the order of the distribution of dimensions in the environmental system state vector to obtain the adjustment constraint vector for that sampling time. The methods for obtaining the disorder evaluation index include: Based on the distribution of the adjustment constraint vector at the previous time-series upsampling time, the judgment curve at the current time is obtained; based on the distribution trend of the adjustment constraint vector between the initial time and the current time, the error advancement line at the current time is obtained. The shortest distance between the data point corresponding to each sampling moment on the judgment curve and the error advancement line is taken as the deviation distance value at each sampling moment on the judgment curve; the deviation distance value at each sampling moment on the judgment curve is integrated from the initial moment to the current moment to obtain the limit range value at the current moment; Multiply the time difference between the initial time and the current time by the mean of all deviation distance values ​​in the judgment curve to obtain the error change value at the current time. The difference between the current limit range value and the error change value is used as the disorder evaluation index at the current moment. The method for obtaining the adjusted sampling frequency includes: The product of the current sampling frequency and the disorder adjustment coefficient is used as the frequency adjustment value at the current time; the sum of the current sampling frequency and the frequency adjustment value is used as the initial adjustment frequency at the current time. When the initial adjustment frequency is greater than or equal to the preset lower frequency limit and less than or equal to the preset upper frequency limit, the initial adjustment frequency is used as the adjustment sampling frequency at the current sampling time. When the initial adjustment frequency is less than the preset lower limit, the preset lower limit will be used as the adjustment sampling frequency at the current sampling time. When the initial adjustment frequency is greater than the preset frequency upper limit, the preset frequency upper limit will be used as the adjustment sampling frequency at the current sampling time.

2. The propulsion control method for an underwater robot according to claim 1, characterized in that, The method for obtaining the limit value includes: For any dimension at the sampling time, calculate the mean of the values ​​of all elements in that dimension at the sigma point at that sampling time, and use it as the distribution mean of that dimension; calculate the standard deviation of the values ​​of all elements in that dimension at the sigma point at that sampling time, and use it as the distribution deviation of that dimension. Subtract the mean of the distribution from the value of each element in the dimension at the sigma point at the sampling time, and then divide by the distribution deviation value to obtain the standardized value of each element in the dimension. After calculating the fourth power of the standardized value of each element in this dimension, the cumulative value is calculated to obtain the constraint value of this dimension.

3. The propulsion control method for an underwater robot according to claim 1, characterized in that, The method for obtaining the judgment curve includes: Each element in the adjustment constraint vector is used as a dimension to establish a multidimensional spatial coordinate system. The adjustment constraint vector at each sampling time is then mapped to the multidimensional spatial coordinate system to obtain the data points at each sampling time in the multidimensional spatial coordinate system. Connect the data points in chronological order of sampling time to obtain the judgment curve.

4. The propulsion control method for an underwater robot according to claim 3, characterized in that, The method for obtaining the error propagation line includes: The line connecting the data points corresponding to the initial time and the current time in the multidimensional coordinate system is used as the error propagation line.

5. The propulsion control method for an underwater robot according to claim 1, characterized in that, The method for obtaining the adjusted switching gain includes: The sum of the cosine similarity of the adjustment constraint vector between the initial time and the current time and the preset adjustment value is used as the stability adjustment coefficient at the current time. The disorder evaluation index at the current moment is normalized and range-mapped to obtain the disorder adjustment coefficient at the current moment. The product of the switching gain and the current chaos adjustment coefficient is used as the current gain adjustment value; the sum of the current switching gain and the gain adjustment value is used as the current initial adjustment gain. The product of the initial adjustment gain and the stable adjustment coefficient at the current moment is used as the adjustment switching gain at the current moment.

6. A propulsion control system for an underwater robot, characterized in that, The system includes: The data acquisition module is used to acquire the environmental system state vector of the underwater robot at each sampling time from the initial time to the current time; and to perform unscented Kalman filtering on the environmental system state vector at each sampling time to obtain the sigma point at the sigma point generation stage at each sampling time, as well as the predicted state vector at the current time. The adjustment constraint analysis module is used to obtain the adjustment constraint vector for each sampling time based on the fluctuation of the numerical distribution in each dimension of the sigma point at each sampling time; the method for obtaining the adjustment constraint vector includes: For any given sampling time, based on the degree of concentration of the values ​​in each dimension of the sigma point, the constraint value for each dimension is obtained; the constraint values ​​for each dimension at that sampling time are arranged in the order of the distribution of dimensions in the environmental system state vector to obtain the adjustment constraint vector for that sampling time. The trend disorder analysis module is used to obtain a disorder evaluation index for the current moment based on the distribution deviation between the trend of the adjustment constraint vector changes of all continuously distributed sampling moments before the current moment and the overall trend of the adjustment constraint vector from the initial moment to the current moment; the method for obtaining the disorder evaluation index includes: Based on the distribution of the adjustment constraint vector at the previous time-series upsampling time, the judgment curve at the current time is obtained; based on the distribution trend of the adjustment constraint vector between the initial time and the current time, the error advancement line at the current time is obtained. The shortest distance between the data point corresponding to each sampling moment on the judgment curve and the error advancement line is taken as the deviation distance value at each sampling moment on the judgment curve; the deviation distance value at each sampling moment on the judgment curve is integrated from the initial moment to the current moment to obtain the limit range value at the current moment; Multiply the time difference between the initial time and the current time by the mean of all deviation distance values ​​in the judgment curve to obtain the error change value at the current time. The difference between the current limit range value and the error change value is used as the disorder evaluation index at the current moment. The switching gain adjustment module is used to obtain the switching gain at the current moment through the sliding mode controller based on the current environmental system state vector and the predicted state vector; and to obtain the adjusted switching gain at the current moment based on the cosine similarity of the adjustment constraint vector between the current moment and the initial moment, combined with the disorder evaluation index and the switching gain at the current moment. A control signal output module is used to adjust the sampling frequency at the current moment based on the disorder evaluation index at the current moment, thereby obtaining the adjusted sampling frequency at the current sampling moment; based on the adjusted sampling frequency at the current moment, the adjusted switching gain is re-input into the sliding mode controller, and the control signal for the next moment is output; the method for obtaining the adjusted sampling frequency includes: The product of the current sampling frequency and the disorder adjustment coefficient is used as the frequency adjustment value at the current time; the sum of the current sampling frequency and the frequency adjustment value is used as the initial adjustment frequency at the current time. When the initial adjustment frequency is greater than or equal to the preset lower frequency limit and less than or equal to the preset upper frequency limit, the initial adjustment frequency is used as the adjustment sampling frequency at the current sampling time. When the initial adjustment frequency is less than the preset lower limit, the preset lower limit will be used as the adjustment sampling frequency at the current sampling time. When the initial adjustment frequency is greater than the preset frequency upper limit, the preset frequency upper limit will be used as the adjustment sampling frequency at the current sampling time.

7. A propulsion control device for an underwater robot, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a propulsion control method for an underwater robot as described in any one of claims 1 to 5.

Images