Closed-loop feedback control method and system for a marine wave-compensated medical platform

Abstract

The present application relates to the technical field of ship wave compensation, in particular to a closed-loop feedback control method and system for a marine wave compensation medical platform, the method comprising the following steps: acquiring the attitude signal of the wave compensation medical platform and constructing a data set, and then constructing a lower platform pose prediction model to obtain the attitude prediction value at the future time; based on the data set and the lower platform pose prediction model, a residual prediction model is constructed to obtain the residual prediction value of the attitude prediction; a kinematics model of the lower platform is constructed, the attitude accurate value is determined through the attitude correction algorithm combined with the attitude prediction value and the residual prediction value, and then the motion back solution algorithm is used to solve the rotation angle of the servo motor, and the closed-loop feedback control of the wave compensation medical platform is realized. The present application can realize accurate control of the wave compensation medical platform in complex environment, improve the wave compensation effect of the wave compensation medical platform, and thus reduce the risk of offshore medical treatment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ship wave compensation technology, in particular to a closed-loop feedback control method and system for a ship wave compensation medical treatment platform. BACKGROUND

[0002] During a long sea voyage, especially when encountering complex sea conditions or performing emergency rescue tasks, a stable surgical environment is crucial to ensure the success rate of medical treatment and the safety of patients. The sway of the ship not only affects the daily rest of the crew, but for the medical team performing delicate operations on the ship, any slight sway can interfere with the accuracy of the operation and increase the risk of surgery. Therefore, the prior art proposes a ship medical treatment bed with wave compensation function. In addition to including an upper platform and a lower platform, such a medical treatment bed usually includes an adjusting device for adjusting the attitude of the upper platform, such as a servo motor, an electric cylinder, a connecting rod, etc.

[0003] However, although the existing wave compensation ship medical treatment bed adjusts the attitude of the upper platform through devices such as servo motors, electric cylinders, connecting rods, etc., its control method is still relatively rough and cannot meet the high-precision surgery requirements, becoming a bottleneck restricting the quality of maritime medical treatment. Among them, some wave compensation ship medical treatment beds use an open-loop control system, which can only roughly adjust the actuator action according to the current sway data of the ship body, and the compensation accuracy is difficult to control. Others use a pure feedback closed-loop control system, which can achieve high-precision wave compensation in regular wave sea conditions, but in irregular wave sea conditions, the compensation effect may be limited due to the randomness of sea wave changes. SUMMARY

[0004] In view of the defects in the prior art, the present application provides a closed-loop feedback control method and system for a wave compensation medical treatment platform on a ship.

[0005] To achieve the above-mentioned purpose, in a first aspect, the present application provides a closed-loop feedback control method for a wave compensation medical treatment platform on a ship, comprising the following steps: acquiring the attitude signal of the lower platform in the wave compensation medical treatment platform in real time, and using the attitude signal to establish a data set, and then constructing a lower platform pose prediction model; based on the data set and the lower platform pose prediction model, constructing a residual prediction model; using the lower platform pose prediction model to obtain the attitude prediction value at the future time, and using the residual prediction model to obtain the residual prediction value of the attitude prediction; constructing the kinematics model of the lower platform, combining the attitude prediction value and the residual prediction value, and determining the attitude accurate value through an attitude correction algorithm; according to the attitude accurate value, using a motion inverse algorithm to solve the rotation angle of the servo motor, and then realizing the closed-loop feedback control of the wave compensation medical treatment platform. The present application can realize accurate control of the wave compensation medical treatment platform in complex environments, improve the wave compensation effect of the wave compensation medical treatment platform, and thus reduce the risk of maritime medical treatment.

[0006] Optionally, the real-time acquisition of the attitude signal of the lower platform in the wave-compensated medical care platform is performed, and the attitude signal is used to establish a data set, and then a lower platform pose prediction model is constructed, including the following steps: real-time acquisition of the rotation angle of the lower platform around the X-axis and the Y-axis and the vertical displacement in the Z-axis direction; the rotation angle and the vertical displacement collected at the same time are taken as a sample data, and after a certain number of sample data are collected, a data set is constructed using all the sample data; according to the data set, a plurality of lower platform pose prediction models are constructed using an autoregressive model, which are respectively used to predict the rotation angle and the vertical displacement. This method constructs a data set by collecting the attitude signal of the lower platform in the environment of the ship, and then constructs a lower platform pose prediction model using an autoregressive model, so that the model has pertinence and real-time performance.

[0007] Optionally, the recursive least squares method with a forgetting factor is used to realize online identification of the parameters in the autoregressive model, and the forgetting factor is adjusted in real time through Bayesian optimization. This method uses the recursive least squares method with a forgetting factor for online parameter identification, so that the model can continuously update its parameters using the latest data, thereby overcoming the model misalignment problem caused by changes in the dynamic characteristics of the sea waves; in addition, the forgetting factor is adjusted in real time through Bayesian optimization, realizing the intelligentization of the parameter update strategy, and improving the robustness and accuracy of the model in complex time-varying marine environments.

[0008] Optionally, based on the data set and the lower platform pose prediction model, a residual prediction model is constructed, including the following steps: based on the data set and the lower platform pose prediction model, a prediction error data set is constructed; according to the prediction error data set, a lightweight neural network is used to dynamically model error transmission to obtain a residual prediction model. This method uses a lightweight neural network to dynamically model error transmission, which is different from simple static error correction. It can learn and capture the complex and nonlinear dynamic trends of errors, and then intelligently predict the possible deviations of the predicted values at future time points, thereby providing a high-precision and forward-looking error estimate value for subsequent filtering algorithms, which helps to achieve accurate wave compensation.

[0009] Optionally, the lower platform pose prediction model is used to obtain the attitude prediction value at the future time, and the residual prediction model is used to obtain the residual prediction value of the attitude prediction, including the following steps: the lower platform pose prediction model is used to obtain the attitude prediction value at the future time, and the attitude prediction value includes a rotation angle prediction value and a vertical displacement prediction value; the residual prediction model is used to obtain the residual prediction value of the rotation angle prediction value and the vertical displacement prediction value.

[0010] Optionally, the method further comprises the following steps of: constructing the kinematic model of the lower platform; using the extended Kalman filter to obtain the accurate attitude value according to the kinematic model, the predicted attitude value and the predicted residual value.

[0011] Optionally, the kinematic model comprises a roll model, a pitch model and a heave model.

[0012] Optionally, the method of using the extended Kalman filter to obtain the accurate attitude value according to the kinematic model, the predicted attitude value and the predicted residual value comprises the following steps of: taking the kinematic model as a state transition equation, and taking the predicted attitude value and the predicted residual value as observation inputs to generate an initial state and an observation vector of the extended Kalman filter; and outputting the accurate attitude value fused with the model prediction and the residual compensation through a prediction-update iteration process of the extended Kalman filter based on the initial state and the observation vector. This method dynamically weighs the reliability of the calculation result of the physical model and the prediction result of the data-driven model by taking the kinematic model as the state transition equation of the extended Kalman filter and taking the predicted attitude and the predicted residual into the observation vector, and finally outputs the accurate attitude value, which provides a crucial reliable input for the motion back analysis and control, and fundamentally improves the stability and compensation performance of the entire wave compensation system.

[0013] Optionally, the method of using the motion back analysis algorithm to solve the rotation angle of the servo motor according to the accurate attitude value to realize the closed-loop feedback control of the wave compensation medical platform comprises the following steps of: using the motion back analysis algorithm to solve the rotation angle of the servo motor based on the accurate attitude value; driving the servo motor to rotate using the servo driver according to the rotation angle of the servo motor; and using the encoder to collect the actual rotation angle of the servo motor in real time and feed back to the servo driver, thereby continuously adjusting the motor action to keep the upper platform of the wave compensation medical platform horizontal. This method dynamically maintains the stability of the upper platform by using the encoder to detect the actual rotation angle in real time to form a negative feedback loop, thereby achieving the ultimate goal of active wave compensation and improving the operation safety and reliability of the wave compensation medical platform in a turbulent environment.

[0014] In a second aspect, the present application provides a closed-loop feedback control system of a wave-compensated medical platform for ships, which is configured to use the closed-loop feedback control method of a wave-compensated medical platform for ships provided by the present application, and comprises: a data acquisition device, configured to acquire a posture signal of a lower platform in the wave-compensated medical platform in real time; a control center, configured to use the posture signal to build a data set, and then build a lower platform pose prediction model; build a residual prediction model based on the data set and the lower platform pose prediction model; use the lower platform pose prediction model to acquire a posture prediction value at a future time, and use the residual prediction model to acquire a residual prediction value of the posture prediction; build a kinematics model of the lower platform, combine the posture prediction value and the residual prediction value, and determine a posture accurate value through a posture correction algorithm; use a motion back-solving algorithm to solve a rotation angle of a servo motor according to the posture accurate value, and then realize closed-loop feedback control of the wave-compensated medical platform; a wave compensation execution device, comprising a servo driver and a servo motor, configured to respond to a control instruction of the control center, and then realize wave compensation of the wave-compensated medical platform; and a control handle, configured to control start and stop of the wave compensation. The system can improve the practicability of the method and facilitate the popularization of the method. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0016] Figure 1 A flowchart of a closed-loop feedback control method of a wave-compensated medical platform for ships according to an embodiment of the present application;

[0017] Figure 2 A roll random signal according to an embodiment of the present application;

[0018] Figure 3 A pitch random signal according to an embodiment of the present application;

[0019] Figure 4 Anti-rolling effect of a simulation prototype on roll according to an embodiment of the present application;

[0020] Figure 5 Anti-rolling effect of a simulation prototype on pitch according to an embodiment of the present application;

[0021] Figure 6A schematic diagram of a framework of a closed-loop feedback control system of a marine wave-compensated medical treatment platform according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Specific embodiments of the present application will now be described in detail with reference to the following figures. It should be appreciated that the embodiments described herein are only given by way of example and are not used to limit the present application. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, it will be apparent to one of ordinary skill in the art that the present application can be practiced without these specific details. In other instances, well-known circuits, software or methods have not been described in detail in order to avoid obscuring the present application.

[0023] Throughout this specification, the use of "one embodiment", "an embodiment", "one example", or "an example" means that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. Therefore, appearances of the phrases "in one embodiment", "in an embodiment", "one example", or "an example" in various places throughout this specification are not necessarily referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics can be combined in any suitable

[0024] It should be noted that in an alternative embodiment, the same symbols or letters appearing in all formulas represent the same meaning and values except for the independent description.

[0025] In an alternative embodiment, referring to Figure 1 The present application provides a closed-loop feedback control method of a marine wave-compensated medical treatment platform, the method comprising the following steps:

[0026] S1, acquiring a posture signal of a lower platform in the wave-compensated medical treatment platform in real time, and using the posture signal to establish a data set, and then constructing a lower platform pose prediction model.

[0027] The step S1 specifically comprises the following steps:

[0028] S11, collecting a rotation angle of the lower platform around the X-axis and the Y-axis and a vertical displacement in the Z-axis direction in real time.

[0029] Specifically, in the embodiment, the wave-compensated medical treatment platform, i.e., a medical treatment bed with a wave-compensation function on a ship, comprises an upper platform (bed surface) and a lower platform (bed bottom) capable of being fixed on the ship, and a servo motor, an electric cylinder, a connecting rod and other devices for wave compensation. The attitude signal of the lower platform comprises the rotation angles of the lower platform around the X-axis and the Y-axis and the vertical displacement of the lower platform in the Z-axis direction.

[0030] In the embodiment, the rotation angles of the ship around the X-axis and the Y-axis and the heave displacement are obtained by using a motion reference unit (MRU) installed on the ship. Since the lower platform is rigidly connected with the ship body when the lower platform is fixed on the ship, the rotation angles of the ship around the X-axis and the Y-axis can be directly taken as the rotation angles of the lower platform around the X-axis and the Y-axis, and then the vertical displacement of the lower platform in the Z-axis direction is calculated according to the following relationship:

[0031]

[0032] wherein, is the vertical displacement of the lower platform in the Z-axis direction, is the heave displacement of the ship, is the lateral distance of the geometric center of the lower platform relative to the center of gravity of the ship, is the longitudinal distance of the geometric center of the lower platform relative to the center of gravity of the ship. The method for determining the position of the center of gravity of the ship can refer to the prior art means.

[0033] In other alternative embodiments, a 2-axis tilt sensor can also be used to measure the rotation angles of the lower platform around the X-axis and the Y-axis, and an inertial measurement unit (IMU) can be used to measure the vertical displacement of the lower platform in the Z-axis direction.

[0034] S12, the rotation angle and the vertical displacement collected at the same time are taken as a sample data, and after a set number of sample data are collected, all the sample data are used to construct a data set.

[0035] Specifically, in the embodiment, the samples in the data set are arranged in the time dimension in the order of sequence. The number of samples in the data set needs to be determined according to the dominant wave period of the target sea area and the sampling frequency of the sensor, and the value is not less than 5 times the number of data points corresponding to the dominant wave period, preferably 10 to 20 times. For example, for a system with a dominant wave period of 10 seconds and a sampling frequency of 50 Hz, 5000 to 10000 sample data are preferred.

[0036] S13, according to the data set, a plurality of lower platform pose prediction models are constructed using an autoregressive model, which are respectively used to predict the rotation angles and the vertical displacement.

[0037] Specifically, in the embodiment, the rotation angles of the lower platform around the X-axis and the Y-axis are predicted by using the autoregressive model. and and the vertical displacement of the lower platform in the Z-axis direction The first model, the second model and the third model are respectively constructed by using an autoregressive model, and are used for predicting the rotation angle of the lower platform around the X-axis, the rotation angle of the lower platform around the Y-axis and the vertical displacement of the lower platform in the Z-axis direction.

[0038] When the lower platform pose prediction model is constructed, the order of the autoregressive model is determined by balancing the model fitting degree and complexity by minimizing the information criterion.

[0039]

[0040] wherein, is the data collected by the sensor at the k moment, p is the order of the autoregressive model, is the i-th to-be-identified parameter, is the data collected by the sensor at the k-i moment, c is a steady-state component, is white noise.

[0041] More specifically, the recursive least squares (RLS) with a forgetting factor is used to realize online identification of the parameters in the autoregressive model, and the forgetting factor is adjusted in real time through Bayesian optimization. The specific process is as follows:

[0042] 1. After determining the order of the autoregressive model, the parameter estimation and the covariance matrix of the RLS are initialized, and the search space of the forgetting factor is set, and a value in the search space is randomly selected as the initial value of the forgetting factor. The search space of the forgetting factor is [0.9, 0.999], and the initialized parameter estimation and the covariance matrix satisfy the following relationship:

[0043]

[0044]

[0045] wherein, is the initialized parameter, which is a vector composed of the to-be-identified parameters in the autoregressive model, is the initialized covariance matrix; is a very small positive number, ; I is the unit matrix.

[0046] 2. For , or , the k moment is taken as the current moment, when the sensor collects the data of the current moment, the RLS updates the regression vector, the gain vector, the parameter estimation and the covariance matrix, that is:

[0047]

[0048]

[0049]

[0050]

[0051] wherein, is the updated regression vector, that is, the regression vector of the current moment; is the data collected by the sensor at the k-i moment, n is the length of the regression vector; T is the transpose; is the updated gain vector, that is, the gain vector of the current moment; is the covariance matrix at the k-1 moment; is the forgetting factor; is the updated parameter, that is, the parameter estimation of the current moment; is the parameter at the k-1 moment; is the data collected by the sensor at the k moment, that is, the data of the current moment; is the updated covariance matrix, that is, the covariance matrix of the current moment.

[0052] 3, the updated parameter is brought into the next platform pose prediction model, the predicted value at the k+1 moment is calculated, and the prediction error is further calculated, for the prediction error of the k+1 moment , which satisfies , is the predicted value at the k+1 moment.

[0053] 4, the mean square error MSE of the prediction error of the last m moments under the current forgetting factor is calculated, and when the MSE is greater than the mean square error threshold, the forgetting factor that minimizes the MSE is searched in the search space through Bayesian optimization with the objective of minimizing the MSE, and is recorded as the optimal forgetting factor.

[0054] 5, the optimal forgetting factor is used to replace the original forgetting factor in the RLS.

[0055] This embodiment constructs a dataset by collecting attitude signals from the platform under the ship's operating environment. An autoregressive model is then used to build a platform pose prediction model, making the model both targeted and real-time. Specifically, this method employs recursive least squares with a forgetting factor for online parameter identification, enabling the model to continuously update its parameters using the latest data. This overcomes model inaccuracies caused by dynamic changes in ocean waves. Furthermore, Bayesian optimization is used to adjust the forgetting factor in real time, achieving intelligent parameter update strategies and improving the model's robustness and accuracy in complex, time-varying marine environments.

[0056] S2. Based on the dataset and the lower platform pose prediction model, construct a residual prediction model.

[0057] Step S2 specifically includes the following steps:

[0058] S21. Construct a prediction error dataset based on the dataset and the lower platform pose prediction model.

[0059] Specifically, in this embodiment, the data in the dataset is used as the true values. The predicted values ​​corresponding to the true values ​​can be obtained through the first, second, and third models. The corresponding prediction error can then be calculated, and a prediction error dataset can be further constructed. It should be noted that the prediction error dataset does not include the data from the first n samples in the dataset. , and The prediction error is due to the length of the regression vector being n in step S13.

[0060] S22. Based on the prediction error dataset, use a lightweight neural network to dynamically model the error propagation and obtain the residual prediction model.

[0061] Specifically, in this embodiment, the data in the prediction error dataset is sorted chronologically. The first 80% of the data in the prediction error dataset is used as the training set, and the last 20% is used as the validation set to train and validate the lightweight neural network, resulting in a residual prediction model. The input to the lightweight neural network is the current... , and The prediction error is output as the next time step. , and To mitigate prediction errors, the network structure employs a single hidden layer design, containing 16 neurons and using the ReLU activation function to enhance nonlinear fitting capabilities. The output layer uses a linear activation function to adapt to the characteristics of the regression task. The model training uses mean squared error as the loss function to quantify the prediction error, and the Adam optimizer is used for parameter optimization to achieve efficient convergence. During training, the batch size is set to 32 to balance memory efficiency and gradient stability. An early stopping mechanism is used to monitor the validation set loss to prevent overfitting. Finally, the root mean square error is used as the evaluation metric on the validation set to verify the model's ability to capture the spatiotemporal evolution of residuals, thereby ensuring the effectiveness and robustness of the residual prediction model in ship kinematics error compensation scenarios.

[0062] This embodiment utilizes a lightweight neural network to dynamically model the error propagation law, which differs from simple static error correction. It can learn and capture the complex and nonlinear dynamic changes in the error, and then intelligently predict the possible deviations in the predicted value at future moments. This provides a high-precision and forward-looking error estimate for subsequent filtering algorithms, which helps to achieve accurate wave compensation.

[0063] S3. Use the lower platform pose prediction model to obtain the pose prediction value at future time, and use the residual prediction model to obtain the residual prediction value of the pose prediction.

[0064] Step S3 specifically includes the following steps:

[0065] S31. Obtain the attitude prediction value at a future time using the lower platform pose prediction model. The attitude prediction value includes the rotation angle prediction value and the vertical displacement prediction value.

[0066] Specifically, in this embodiment, the lower platform pose prediction model is used to obtain the predicted pose value of the lower platform at future times. Specifically, the first model is used to obtain the predicted rotation angle of the lower platform around the X-axis. The second model is used to obtain the predicted rotation angle of the lower platform around the Y-axis. The third model is used to obtain the predicted vertical displacement of the lower platform in the Z-axis direction. .

[0067] S32. Use the residual prediction model to obtain the residual prediction values ​​of the rotation angle prediction value and the vertical displacement prediction value.

[0068] Specifically, in this embodiment, the current time is used to... , and By inputting the prediction error into the residual prediction model, the prediction for future time intervals can be obtained. , and The predicted residual values.

[0069] S4, constructing a kinematic model of the lower platform, combining the attitude prediction value and the residual prediction value, and determining an attitude accurate value through an attitude correction algorithm.

[0070] The step S4 specifically comprises the following steps:

[0071] S41, constructing a kinematic model of the lower platform.

[0072] Specifically, in the embodiment, since the lower platform is fixed on the ship, the lower platform and the ship body can be regarded as rigidly connected, if only considering the roll, pitch and heave and ignoring the coupling influence of other degrees of freedom, at this time, the rotation angle of the lower platform around the X axis is the roll angle of the ship, the rotation angle of the lower platform around the Y axis is the pitch angle of the ship, and the vertical displacement of the lower platform in the Z axis direction is the vertical displacement of the ship in the Z axis direction plus the additional displacement caused by the rotation movement, then the kinematic model of the lower platform can be obtained by referring to the kinematic model of the ship, including the roll model, the pitch model and the heave model.

[0073] In the embodiment, the ship angle response data is obtained by numerically solving a second-order nonlinear stochastic differential equation, and the dynamic model is constructed based on the rigid body rotation motion equation, that is, the roll model and the pitch model satisfy the following relationships in turn:

[0074]

[0075]

[0076] wherein, is the moment of inertia around the X axis, is the roll angular acceleration, is the linear damping coefficient, is the roll angular velocity, is the roll restoring moment coefficient, is the roll angle, is the external excitation moment in the X axis direction, is the random disturbance moment in the X axis direction, is the moment of inertia around the Y axis, is the pitch angular acceleration, is the pitch damping coefficient, is the pitch angular velocity, is the pitch restoring moment coefficient, is the pitch angle, is the external excitation moment in the Y axis direction, is the random disturbance moment in the Y axis direction.

[0077] The heave model should describe the linear motion of the ship in the vertical direction, which satisfies the following relationship:

[0078]

[0079] where M is the mass of the ship, is the heave acceleration of the ship, is the heave damping coefficient, is the heave velocity of the ship, is the heave restoring force coefficient, is the external excitation force in the Z-axis direction, is the random disturbance force in the Z-axis direction.

[0080] Further, the vertical displacement of the lower platform in the Z-axis direction can be expressed using the relationship of step S11, and the acceleration of the lower platform in the Z-axis direction can be obtained by differentiating the velocity of the lower platform in the Z-axis direction , can be obtained by differentiating .

[0081] When solving the kinematic model of the ship, it can be converted into a first-order differential equation system, and then solved using the fourth-order Runge-Kutta method. When solving, the time step is set to 0.02s, the total time is set to 500s, and the initial conditions are set according to the actual motion state of the ship, such as setting the initial condition to 0 when the ship is stationary. The random component of each time step is independently generated by the randn function to ensure statistical independence.

[0082] S42, according to the kinematic model, the attitude prediction value and the residual prediction value, using extended Kalman filtering to obtain the accurate attitude value.

[0083] Specifically, step S42 further includes the following steps:

[0084] S421, the kinematic model is used as a state transition equation, and the attitude prediction value and the residual prediction value are used as observation inputs to generate the initial state and observation vector of the extended Kalman filter.

[0085] Specifically, in this embodiment, the state variable h of the extended Kalman filter is defined as:

[0086]

[0087] where, is the first derivative of , is the first derivative of .

[0088] The observation vector r of the extended Kalman filter satisfies:

[0089]

[0090] wherein, 、 and are the corresponding residual prediction values. 、 and

[0091] S422, outputting the attitude accurate value fused with the model prediction and the residual compensation based on the initial state and the observation vector through the prediction-update iteration process of the extended Kalman filter.

[0092] Specifically, in the present embodiment, this step is a prior art means, which is not described in detail here.

[0093] The present embodiment establishes the kinematic model as the state transition equation of the extended Kalman filter, and jointly incorporates the attitude and residual prediction into the observation vector, thereby dynamically weighing the reliability of the calculation results of the physical model and the prediction results of the data-driven model, and finally outputting the attitude accurate value, which provides a crucial reliable input for the motion back analysis and control, and fundamentally improves the stability and compensation performance of the entire wave compensation system.

[0094] S5, solving the rotation angle of the servo motor using a motion back analysis algorithm according to the attitude accurate value, and thereby realizing the closed-loop feedback control of the wave compensation medical platform.

[0095] wherein, step S5 specifically comprises the following steps:

[0096] S51, solving the rotation angle of the servo motor using a motion back analysis algorithm based on the attitude accurate value.

[0097] Specifically, in the present embodiment, since the purpose of wave compensation is to keep the upper platform level, the pose of the upper platform is known, and therefore after obtaining the attitude accurate value of the lower platform, the rotation angle of the servo motor is solved through the motion back analysis algorithm. The specific motion back analysis algorithm needs to be determined according to the structure of the wave compensation medical platform. Taking a 3-RPS parallel mechanism as an example, the back analysis calculation principle is as follows:

[0098] 1. mapping the hinge point of the upper platform to the base coordinate system through the homogeneous transformation matrix.

[0099]

[0100] wherein, is the homogeneous transformation matrix; R is the rotation matrix, , represents the rotation matrix (i.e. unit matrix) rotating 0 radian around the Z axis, ​Indicates rotation about the Y-axis The rotation matrix, , Indicates rotation about the X-axis The rotation matrix, w is the translation vector. , It is a fixed offset.

[0101] The coordinate transformation formula is:

[0102]

[0103] in, Let J be the coordinates of the j-th hinge point on the upper platform in the base coordinate system. Let be the coordinates of the j-th hinge point on the upper platform in the local coordinate system. The base coordinate system is a spatial rectangular coordinate system with the center point of the lower platform as the origin, and the local coordinate system is a spatial rectangular coordinate system with the center point of the upper platform as the origin.

[0104] 2. Establish distance constraint equations for each branch.

[0105] Since the length of the branch is fixed, the distances from the center of the upper ball joint to the middle hinge point and the distances from the center of the lower ball joint to the middle hinge point are constants. Therefore, the following distance constraint equation is established for each branch:

[0106]

[0107] in, Let be the coordinates of the intermediate hinge point of the j-th branch. This is the distance from the center of the upper ball joint to the middle hinge point. This is the distance from the center of the lower ball joint to the intermediate hinge point. Let be the coordinates of the j-th hinge point on the lower platform. The Levenberg-Marquardt algorithm is used to solve the above nonlinear equations to obtain the intermediate hinge point of the j-th branch. The coordinates.

[0108] 3. Calculate the angle between the swing arm and the vertical direction based on the coordinates of the intermediate hinge point, and obtain the deviation relative to the initial angle, i.e., the driving joint rotation angle.

[0109]

[0110]

[0111] in, Let be the angle between the j-th branch and the vertical direction. From the origin O to The vector, Let be the unit vector along the Z-axis in the base coordinate system. to drive the joint rotation angle, to drive the initial angle of the swing arm.

[0112] After obtaining the driving joint rotation angle, the servo motor rotation angle can be determined according to the ratio of the servo motor rotation angle and the driving joint rotation angle, so as to realize the control of the servo motor to complete the wave compensation function.

[0113] S52, according to the servo motor rotation angle, using a servo driver to drive the servo motor to rotate.

[0114] Specifically, in this embodiment, the servo driver adopts a PID control algorithm, which controls the servo motor to rotate according to the servo motor rotation angle after obtaining the servo motor rotation angle.

[0115] S53, using an encoder to collect the actual rotation angle of the servo motor in real time and feed back to the servo driver, and then continuously adjusting the motor action to keep the upper platform of the wave compensation medical platform horizontal.

[0116] This embodiment forms a negative feedback loop by real-time detection of the actual rotation angle by the encoder, thereby dynamically maintaining the stability of the upper platform, achieving the ultimate goal of active wave compensation, and improving the operation safety and reliability of the wave compensation medical platform in a turbulent environment.

[0117] To verify the performance of this scheme, this embodiment uses a ship active compensation medical bed disclosed in patent number CN222584855U to perform a simulation test of sea wave compensation.

[0118] A simulation machine for simulation test is established in the MATLAB environment, taking ship roll and pitch as an example, to show the anti-swing ability of the simulation machine under this scheme. When performing the simulation experiment, the corresponding random swing signals are generated according to the roll model and the pitch model, and the corresponding parameter settings are: t is time, is a Gaussian white noise with a mean of 0 and a standard deviation of 6000, is a Gaussian white noise with a mean of 0 and a standard deviation of 6000.

[0119] The generated random swing signals are shown in Figure 2 and Figure 3 . Among them, Figure 2 (a) in (a) is the change of roll angle with time, Figure 2 (b) in (b) is the change of roll angular velocity with time,​​​​​​Figure 3 In the figure, (a) represents the change of pitch angle over time. Figure 3 (b) in the figure represents the change of pitch angular velocity over time.

[0120] The simulation prototype's anti-sway effect against roll is as follows: Figure 4 As shown, the anti-sway effect against pitch is as follows: Figure 5 As shown. From Figure 4 and Figure 5 As can be seen, wave compensation begins between 100 and 120 seconds. Before wave compensation begins, there is a data collection and model training phase. The upper and lower platforms exhibit the same swaying pattern. However, after wave compensation begins, the lower platform continues to sway, while the upper platform can maintain a relatively stable position, proving the feasibility of this solution.

[0121] It should be noted that in some cases, the actions described in the specification can be performed in different orders and still achieve the desired results. In this embodiment, the order of steps is given only to make the embodiment clearer and easier to explain, and not to limit it.

[0122] In one optional embodiment, please refer to Figure 6 To improve the practicality and facilitate the promotion of this method, the present invention also provides a closed-loop feedback control system for a marine wave compensation medical platform. The closed-loop feedback control system for the marine wave compensation medical platform includes a data acquisition device 1, a control center 2, a wave compensation execution device 3, and a control handle 4. The data acquisition device 1, the control center 2, and the wave compensation execution device 3 transmit signals to each other via a data cable, and the control handle 4 is connected to the control center 2 via Bluetooth.

[0123] The data acquisition device 1 is used to acquire the attitude signal of the lower platform in the wave-compensated medical platform in real time. The data acquisition device 1 includes sensors installed on the wave-compensated medical platform and motion reference units on the ship.

[0124] The control center 2 is used to construct a dataset using the attitude signals, and then construct a lower platform pose prediction model; based on the dataset and the lower platform pose prediction model, a residual prediction model is constructed; the lower platform pose prediction model is used to obtain the attitude prediction value at future time, and the residual prediction model is used to obtain the residual prediction value of the attitude prediction; a kinematic model of the lower platform is constructed, and the attitude prediction value and the residual prediction value are combined to determine the attitude accuracy value through an attitude correction algorithm; based on the attitude accuracy value, a motion inverse algorithm is used to solve the rotation angle of the servo motor, thereby realizing closed-loop feedback control of the wave compensation medical platform.

[0125] The wave compensation execution device 3 comprises a servo driver and a servo motor, and is used for realizing wave compensation of the wave compensation medical platform in response to a control instruction of the control center 2.

[0126] The control handle 4 is used for controlling start and stop of the sea wave compensation.

[0127] The present application has the following beneficial effects:

[0128] Firstly, the method realizes online identification of parameters in the autoregressive model by using a recursive least square method with a forgetting factor, and adjusts the forgetting factor in real time through Bayesian optimization, and then uses the autoregressive model to construct a lower platform pose prediction model, so that the model has pertinence and real-time performance, and also has good robustness and accuracy in a complex time-varying marine environment. Secondly, the method uses a lightweight neural network to dynamically model error transmission rules, intelligently predicts possible deviations of the predicted value at the future time, and then combines the prediction result of the lower platform pose prediction model and the deviation prediction result, uses Kalman filtering to obtain an accurate value of the pose, and provides a crucial reliable input for motion back analysis and control, thereby fundamentally improving the stability and compensation performance of the entire wave compensation system. Finally, the motion back analysis algorithm is used to solve the rotation angle of the servo motor according to the accurate value of the pose, so as to drive the servo motor to rotate by using the servo driver, and the actual rotation angle is detected in real time by using the encoder to form a negative feedback loop, thereby dynamically maintaining the stability of the upper platform, realizing the final purpose of active wave compensation, improving the operation safety and reliability of the wave compensation medical platform in a turbulent environment, and reducing the risk of offshore medical treatment. In addition, the present application provides a system adapted to the method, which helps to improve the practicability of the method and facilitate the popularization of the method.

[0129] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and they should be covered in the scope of the claims and the specification of the present application.

Claims

1. A closed loop feedback control method for a marine wave-compensated medical platform, characterized in that, The method comprises the following steps: Real-time acquisition of the attitude signal of the lower platform in the wave-compensated medical platform, and use of the attitude signal to establish a data set, and then construction of a lower platform pose prediction model; Based on the data set and the lower platform pose prediction model, a residual prediction model is constructed; Using the lower platform pose prediction model to obtain the attitude prediction value at the future time, and using the residual prediction model to obtain the residual prediction value of the attitude prediction; Constructing the kinematic model of the lower platform; Using the kinematic model as the state transition equation, and using the attitude prediction value and the residual prediction value as the observation input, generating the initial state and observation vector of the extended Kalman filter; Based on the initial state and the observation vector, the prediction-update iteration process of the extended Kalman filter is used to output the attitude accurate value of the fusion model prediction and residual compensation; According to the attitude accurate value, the motion back-solving algorithm is used to solve the rotation angle of the servo motor, and then the closed-loop feedback control of the wave-compensated medical platform is realized.

2. The closed loop feedback control method of a wave compensated medical platform for a ship according to claim 1, characterized in that, The real-time acquisition of the attitude signal of the lower platform in the wave-compensated medical platform, and use of the attitude signal to establish a data set, and then construction of a lower platform pose prediction model, comprises the following steps: Real-time acquisition of the rotation angle of the lower platform around the X and Y axes and the vertical displacement in the Z axis direction; The rotation angle and the vertical displacement collected at the same time are used as a sample data, and after a certain number of sample data are collected, all sample data are used to construct a data set; According to the data set, a plurality of lower platform pose prediction models are constructed using an autoregressive model, which are used to predict the rotation angle and the vertical displacement respectively.

3. The closed-loop feedback control method of the wave-compensated medical platform for ships according to claim 2, characterized in that: The recursive least squares method with a forgetting factor is used to realize online identification of the parameters in the autoregressive model, and the forgetting factor is adjusted in real time through Bayesian optimization.

4. The closed loop feedback control method of a wave compensated medical platform for marine use according to claim 1, characterized in that, The construction of the residual prediction model based on the data set and the lower platform pose prediction model comprises the following steps: Based on the data set and the lower platform pose prediction model, a prediction error data set is constructed; According to the prediction error data set, a lightweight neural network is used to dynamically model error transmission to obtain a residual prediction model.

5. The closed loop feedback control method of a wave compensated medical platform for marine use according to claim 1, characterized in that, The use of the lower platform pose prediction model to obtain the attitude prediction value at the future time, and the use of the residual prediction model to obtain the residual prediction value of the attitude prediction, comprises the following steps: The lower platform pose prediction model is used to obtain the attitude prediction value at the future time, which includes the rotation angle prediction value and the vertical displacement prediction value; The residual prediction model is used to obtain the residual prediction value of the rotation angle prediction value and the vertical displacement prediction value.

6. The closed-loop feedback control method of the wave-compensated medical platform for ships according to claim 1, characterized in that: The kinematic model includes a roll model, a pitch model and a heave model.

7. The closed loop feedback control method of a ship-based wave-compensated medical platform according to claim 1, wherein, The use of the motion back-solving algorithm to solve the rotation angle of the servo motor according to the attitude accurate value, and then the closed-loop feedback control of the wave-compensated medical platform, comprises the following steps: Based on the accurate attitude value, a motion inverse solution algorithm is used to solve the rotation angle of the servo motor; According to the rotation angle of the servo motor, a servo driver is used to drive the servo motor to rotate; An encoder is used to collect the actual rotation angle of the servo motor in real time and feed it back to the servo driver, so as to continuously adjust the motor action and keep the upper platform of the wave compensation medical platform horizontal. 8.A closed-loop feedback control system of a ship-based wave-compensated medical platform, configured to use the closed-loop feedback control method of the ship-based wave-compensated medical platform according to any one of claims 1-7, characterized in that, Comprise: A data acquisition device is used to acquire the attitude signal of the lower platform of the wave compensation medical platform in real time; A control center is used to construct a data set using the attitude signal, and then construct a lower platform pose prediction model; based on the data set and the lower platform pose prediction model, a residual prediction model is constructed; the lower platform pose prediction model is used to obtain the attitude prediction value at the future time, and the residual prediction model is used to obtain the residual prediction value of the attitude prediction; a kinematics model of the lower platform is constructed; the kinematics model is used as a state transition equation, and the attitude prediction value and the residual prediction value are used as observation inputs to generate the initial state and observation vector of the extended Kalman filter; Based on the initial state and the observation vector, the prediction-update iteration process of the extended Kalman filter is used to output the attitude accurate value fused with model prediction and residual compensation; based on the attitude accurate value, a motion inverse solution algorithm is used to solve the rotation angle of the servo motor, so as to realize the closed-loop feedback control of the wave compensation medical platform; A wave compensation execution device comprising a servo driver and a servo motor is used to respond to the control instruction of the control center, so as to realize the wave compensation of the wave compensation medical platform; A control handle is used to control the start and stop of the wave compensation.

Images