Fin ray driving type bionic pectoral fin spanwise and chordwise cooperative motion control method and device and medium

Abstract

The invention discloses a fin ray driving type bionic pectoral fin spanwise and chordwise cooperative motion control method and device and a medium, and the method comprises the steps that a bionic pectoral fin is composed of a fin root, a fin ray and a fin surface, the first fin root is subjected to spanwise fluctuation, the second fin root and the third fin root are subjected to spanwise fluctuation and adduction and abduction coupling motion, and the second fin root and the third fin root are subjected to spanwise fluctuation and abduction coupling motion; three-dimensional cooperative movement of the pectoral fins in the spanwise direction, the chordwise direction and the adduction / abduction direction is achieved through the rhythm signal generator; a state sensing module is used for obtaining the motion state of the pectoral fin, an inner-outer nested double-layer optimization strategy is constructed based on a depth deterministic strategy gradient algorithm, an inner layer optimizes local structure rigidity, an outer layer optimizes motion parameters, and a pectoral fin driving mechanism achieves rapid and stable target angle tracking through a trajectory tracking controller. Therefore, a perception-decision-making-control framework is constructed; according to the method, the propelling performance of the robotic fish can be improved in a complex water area, the stability of the robotic fish can be enhanced, and the technical problems of efficient propelling and stable control of the fin ray driving type bionic pectoral fin under the unstructured working condition are solved.

Description

Technical Field

[0001] The present invention relates to the field of bionic robot control, and in particular to a method, device and medium for controlling spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin. Background Art

[0002] In the field of underwater bionic robots, fish pectoral fins are widely considered to be an important propulsion structure for achieving high maneuverability and complex posture adjustment. Compared with traditional propeller drive methods, bionic pectoral fins have lower disturbances, higher fluid adaptability and more degrees of freedom, and therefore have broad application prospects in scenarios such as underwater detection, environmental monitoring and intelligent inspection. However, the currently common bionic pectoral fins mostly use a single fin ray structure with a relatively simple movement mode, which makes it difficult to achieve stable posture control while ensuring propulsion efficiency, especially in unstructured flow fields and complex working conditions.

[0003] Existing research shows that in order to improve the hydrodynamic performance and control ability of pectoral fins, researchers have proposed a multi-fin ray coordinated fluctuation mechanism, that is, through the phase difference between multiple fin rays to achieve spanwise and chordwise fluctuations on the fin surface, thereby improving thrust direction control and attitude response speed; but the current control method still faces the following challenges: (1) How to construct a composite motion model with both spanwise and chordwise fluctuations; (2) How to achieve coordinated control and real-time optimization between multiple fin rays in a complex environment; (3) How to make the control output sufficiently smooth and robust to avoid the impact of signal jumps on the actuator.

[0004] In addition, existing research lacks a modeling and optimization mechanism for the coupling relationship between the stiffness of the bionic pectoral fin structure and the control parameters. As a flexible driving component, the stiffness change of the fin will directly affect the executable feasibility of the motion trajectory and the stability of the force output. Therefore, there is an urgent need for a method that can combine the motion control parameters and the structural stiffness parameters for joint optimization, combined with real-time state perception and intelligent control algorithms, to achieve a more efficient and adaptable bionic pectoral fin motion control solution. Summary of the Invention

[0005] In response to the problems of insufficient pectoral fin control accuracy, low coordination efficiency and unstable structural response in the existing technology, the present invention proposes a method, device and medium for controlling the span-wise and chord-wise coordinated motion of a fin-ray driven bionic pectoral fin. This method achieves efficient coordinated fluctuation of the fin surface in time and space by precisely controlling the span-wise fluctuation, chord-wise phase and inward and outward motion of the fin root, thereby significantly improving the propulsion performance and posture stability of the multi-fin-ray driven pectoral fin, and further improving the swimming performance and environmental adaptability of the entire machine.

[0006] The purpose of the present invention is to provide a method, device and medium for controlling the spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin. The method comprises the following steps: (1) based on the skeletal structure of the pectoral fin of fish, a fin-ray driven bionic pectoral fin consisting of independently driven fin roots, elastic fin rays and fin surfaces is designed, the spanwise fluctuation of the fin surface is formed by the active fluctuation of the fin roots and the passive fluctuation of the fin rays, the fluctuation phase difference between the fin roots forms the chordwise fluctuation of the fin surface, and the deflection phase difference between the fin roots forms the adduction and abduction motion of the fin surface; at the same time, according to the characteristic data of the pectoral fin gait, a fin root kinematic model is constructed to define is the fin root motion control parameter; (2) in order to make the fin root kinematic model continuously output a smooth, continuous and rhythmic control signal, a rhythmic signal generator of an improved central pattern generator (CPGs) is constructed with the Hopf oscillator as the core; the rhythmic signal generator is matched with the input parameters of the fin root kinematic model, and the fin root motion control parameter is changed by changing the rhythmic signal generator. The rhythmic signal generator processes the output of the fin root kinematic model and outputs a smooth fin root angular displacement to achieve the three-dimensional coordinated movement of the fin-driven bionic pectoral fin in the span direction, chord direction, inward and outward directions. The smooth fin root angular displacement is defined as (3) During the movement of the pectoral fins, the pectoral fins are mounted on a six-dimensional force sensor to collect the pectoral fin thrust and torque in real time, and the local fluid load is obtained using a strain gauge array. , use the angle encoder to obtain the actual deflection angular displacement of the fin root during movement , using bending sensors to collect the actual angular displacement of the fin root during movement and the passive angular displacement of the fin ray , through the actual deflection angular displacement of the fin root and the actual fluctuation angular displacement of the fin root and the passive angular displacement of the fin ray The linear displacement and equivalent stiffness of the fin are calculated , the actual angular displacement of the fin root is defined as (4) Based on the deep deterministic policy gradient (DDPG) algorithm, the inner and outer optimization strategies of reinforcement learning are constructed. The outer optimization strategy adopts a weighted combination of maximizing propulsion performance and maximizing posture stability as the optimization goal, and optimizes the optimal fin root motion control parameters. , achieving a balance between propulsion performance and attitude stability, and enhancing the system's adaptability to environmental changes; the inner optimization strategy establishes a stiffness adjustment model based on the structural response characteristics of each fin, and dynamically adjusts the equivalent stiffness through a nested optimization strategy , so that the fin structure response is closer to the optimal dynamic deformation state, thereby improving the structural compliance; (5) In order to ensure accurate tracking of the desired angular displacement of the fin root generated by the outer layer optimization strategy, a trajectory tracking controller based on the model predictive control (MPC) method is introduced. The trajectory tracking controller is constructed on the basis of the outer layer control trajectory and the inner layer stiffness adjustment, and receives the current actual angular displacement of the fin root based on the trajectory prediction and rolling optimization mechanism. Desired angular displacement with the fin root The error between them is used to predict the future state evolution and to obtain the optimal control input by rolling solution. , thereby achieving high-precision, low overshoot and strong robustness trajectory tracking execution.

[0007] As a further technical solution, the fin-ray driven bionic pectoral fin is composed of an independently driven fin root, a passively deformed elastic fin ray, and a passively deformed flexible fin surface, wherein the first fin root realizes spanwise fluctuation, and the second and third fin roots further have the coupled movement capability of adduction and abduction on the basis of spanwise fluctuation. The active fluctuation of the fin root and the passive fluctuation of the fin ray realize spanwise fluctuation of the entire fin surface, the fluctuation phase difference realizes chordwise fluctuation of the entire fin surface, and the deflection phase difference realizes adduction and abduction movement of the entire fin surface.

[0008] The equation of the fin root kinematic model is:

[0009] ;

[0010] ;

[0011] Where, For Moment Root fin root fluctuation angular displacement, For the The amplitude of the root fin root angular displacement, For the The offset of the root fin root fluctuation angular displacement, is the fin root wave motion period ratio, which divides the period into four parts. , is the period of wave motion; For Moment Root fin root deflection angular displacement, For the The amplitude of the angular displacement of the root fin, For the Offset of the root fin deflection angular displacement, is the fin root deflection motion period ratio, which divides the period into two parts. , is the deflection motion period, is time; the fin root motion control parameter , When the value is 1, 2, or 3, the elastic fin ray generates fluctuations under the active drive of the fin root by relying on the elastic properties of its own material, and drives the flexible fin surface to achieve passive spanwise fluctuations.

[0012] As a further technical solution, the rhythmic signal generator is constructed by the improved central pattern generator and the fin root kinematic model. The equation of the rhythmic signal generator is as follows:

[0013] ;

[0014] Where, After smoothing by the rhythm signal generator Output, It's about The output of the rhythm signal generator, After smoothing by the rhythm signal generator Output, It's about The output of the rhythm signal generator, is the response frequency (the larger the faster the response), is the damping ratio (controls the smoothness), is time, the smoothed fin root angular displacement is ;

[0015] The rhythmic signal generator not only has a good approximation capability to the original control signal, but also has the capability to maintain the rhythmicity, smoothness and directivity of the trajectory.

[0016] As a further technical solution, the collection and processing process of the pectoral fin thrust and torque, the local fluid load, the actual fin root fluctuation angular displacement, the passive fin ray fluctuation angular displacement, the actual fin root deflection angular displacement and the fin ray motion linear displacement is as follows: First, the pectoral fin thrust and torque on the pectoral fin surface in the three-axis direction are collected in real time by a six-dimensional force sensor installed above the pectoral fin base, which is recorded as , They are Directional thrust of the pectoral fin surface, Respectively around Moment on the axis of the pectoral fin surface;

[0017] At the same time, strain gauge arrays are arranged at the fin root, the middle section of the fin ray and the end of the fin ray, and the distribution of the local fluid load is measured as follows: , Respectively Root of the root fin, Middle and lower root fin rays The fluid load at the end of the root fin is constructed by quadratic interpolation to obtain the fin fluid pressure function along the length of the fin:

[0018] ;

[0019] Where, For the The length of the fin installed at the root of the root fin, is a constant, which is obtained by the fluid load data at the fin root, the middle of the fin ray and the end of the fin ray. The fluid pressure at the end of the root fin is , No. The fluid moment at the end of the root fin is ;

[0020] Secondly, the actual deflection angular displacement of the fin root is collected using an angle encoder The actual angular displacement of the fin root is obtained by a flexible bending sensor. and the passive angular displacement of the fin ray , combining the geometric model and material response relationship, the linear displacement of the fin tip is calculated as:

[0021] ;

[0022] Where, for Moment The end displacement of the root ray normal, for Moment The tangential displacement of the root ray, For the The length of the fin ray installed at the root of the root fin;

[0023] The equivalent stiffness The calculation formula is

[0024] ;

[0025] Where, For the The equivalent stiffness of the root fin (the larger the root fin, the harder it is to bend). The first Passive oscillation angular displacement of the root fin ray, The values are 1, 2, 3;

[0026] Finally, the above data are uniformly sampled, filtered, normalized and feature fused by the state perception module to form a state vector:

[0027] ;

[0028] Where, ;

[0029] This state vector is used as the environmental observation input of the outer optimization strategy to evaluate the system propulsion performance and attitude stability; at the same time, The data is also synchronously transmitted to the inner layer optimization strategy for dynamic correction of fin ray stiffness.

[0030] As a further technical solution, the outer layer optimization strategy takes the maximization of propulsion performance and posture stability as the outer layer optimization goals, and obtains the optimal fin root motion control parameters through DDPG algorithm learning. ,The superscript “˜” only represents the optimal value of the motion control parameter and does not change the meaning of the letter;

[0031] The equation of the outer optimization objective is:

[0032] ;

[0033] Where, is the effective thrust generated per unit power consumption, For power consumption, represents the attitude disturbance of the pectoral fin in the yaw degree of freedom, represents the attitude disturbance of the pectoral fin in the pitch degree of freedom, represents the variance function, is the weight coefficient, when As it gets smaller, the posture stability gets better;

[0034] The inner layer optimization strategy takes maximizing the structural compliance of the fin root during movement as the inner layer optimization goal while keeping the optimal fin root motion control parameters unchanged, and adjusts the equivalent stiffness of each fin ray. , the actual deformation trajectory under fluid load has good structural compliance;

[0035] The equation for the inner optimization objective is:

[0036] ;

[0037] The first term measures the passive bending response of the fin under the action of the fluid, controlling its degree of compliance, and the second term measures the smoothness of the fin motion curve (limiting frequent or drastic stiffness adjustments to avoid material fatigue or uncontrollable deformation). The equivalent stiffness is Time Passive oscillation angular displacement of the root fin ray, The equivalent stiffness is Time The end displacement of the root ray normal, is the weighting coefficient of the two parts, is the pectoral fin movement cycle, For time;

[0038] The outer layer optimization strategy realizes high-level decision-making and control of the fin root motion behavior goals, and the inner layer optimization strategy ensures the flexibility of the fin ray dynamic response by adjusting the stiffness. The two-layer strategy works synergistically to effectively improve the environmental adaptability, posture stability and propulsion performance of the bionic pectoral fin control system under dynamic conditions.

[0039] As a further technical solution, the trajectory tracking controller uses a model predictive control method to ensure accurate tracking of the desired fin root angular displacement generated by the outer layer optimization strategy. Based on trajectory prediction and rolling optimization mechanism, the expected angular displacement of the fin root is used and the actual angular displacement of the fin root Calculate the displacement error, build a dynamic evolution model within the prediction window, and solve the problem in real time by minimizing the sum of squares of the displacement error and the weighted sum of squares of the control inputs. Output a series of future control inputs and obtain the optimal control input in each control cycle. , so as to achieve accurate tracking of the target trajectory;

[0040] The equation of the trajectory tracking controller is:

[0041] ;

[0042] Where, yes Prediction step length Place Expected angular displacement of the root fin, , The prediction step length Place The actual angular displacement of the root fin, , The prediction step length The first The optimal control input of the root fin root is obtained in each control cycle , the superscript “ˆ” only represents the optimal value and does not change the meaning of the letter; is the error weight matrix, is the control cost matrix, The step length is predicted; the trajectory tracking controller is used to enable the drive mechanism to achieve fast, smooth, and overshoot-free target angle tracking response, thereby constructing a complete perception-decision-control closed-loop path.

[0043] As a further technical solution, the fin-driven bionic pectoral fin is made of rigid fin bones, rigid fin roots, elastic fin rays and flexible fin membranes; the fin bones and fin roots are integrally formed through a 3D printing process and are made of materials with good structural strength and molding precision; the fin rays are made of materials with high elasticity and lightweight properties to ensure good deformation response capabilities under fluid loads; the fin membrane is made of flexible and deformable materials to enhance its adaptability and fluctuation performance to hydrodynamic changes.

[0044] The present invention also provides a fin-ray driven bionic pectoral fin span-wise and chord-wise coordinated motion control device, characterized in that it includes: (1) a drive control module, which is used to output the control signal of the rhythmic signal generator and drive the actuator at the fin root to perform the corresponding motion; (2) a state perception module, which collects the state vector of the first moment of the pectoral fin motion through a sensor, and the state variables include the pectoral fin thrust and torque, the local fluid load, the actual fin root fluctuation angular displacement, the fin ray passive fluctuation angular displacement and the actual fin root deflection angular displacement, which are used for real-time call by the feedback adjustment module; (3) a signal processing module, which generates the fin root motion control parameters through calculation of the outer layer optimization strategy, the inner layer optimization strategy and the trajectory tracking controller; (4) a feedback adjustment module, which compares the data results of the state perception module with the current control parameters, generates dynamic correction information, triggers the outer layer optimization strategy and the inner layer optimization strategy through the interrupt mechanism to re-update the motion strategy, and ensures that the fin root tracks the desired fin root angular displacement through the trajectory tracking controller.

[0045] The present invention also provides a computer-readable storage medium containing motion control program instructions, characterized in that the motion control program instructions are stored in the computer-readable storage medium composed of a Raspberry Pi as a host and a field programmable logic gate array processing unit as a slave, and the computer-readable storage medium is used to implement any of the above-mentioned fin-driven bionic pectoral fin span-wise and chord-wise collaborative motion control methods when executing the motion control program instructions.

[0046] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention.

[0047] The beneficial effects of the present invention are:

[0048] (1) The optimized spanwise and chordwise coordinated motion control method of the fin-ray driven bionic pectoral fin is applicable to a bionic pectoral fin comprising at least three independently drivable fin rays. The first fin ray is subjected to single spanwise fluctuation control. The second fin ray is subjected to spanwise fluctuation and inward and outward movement control, and the spanwise fluctuation has a set phase difference relative to the first fin ray. The third fin ray is subjected to a control strategy similar to that of the second fin ray, and its spanwise fluctuation has a different set phase difference relative to the first and second fin rays. The chordwise fluctuation of the entire fin surface is formed according to the spanwise fluctuation phase difference of the three fin rays.

[0049] (2) In order to achieve a balance between wave performance and structural stability, the three fins are preferably made of highly elastic and lightweight materials, such as thermoplastic polyurethane elastomers or superelastic alloy materials, to ensure sufficient frequency response characteristics and deformation recovery capabilities.

[0050] (3) Based on the gait characteristic data of the pectoral fin, a kinematic model of the fin root is constructed, which makes the pectoral fin have good biomimetic characteristics in terms of movement function. The introduction of a rhythmic signal generator can not only effectively approximate the original control signal, but also maintain the rhythmicity, smoothness and derivability of the output trajectory on this basis, providing a more stable and natural motion drive signal for biomimetic control.

[0051] (4) The outer layer optimization strategy adopts a weighted combination of maximizing propulsion performance and maximizing attitude stability as the optimization goal, optimizes the optimal fin root motion control parameters, achieves a balance between propulsion performance and attitude stability, and enhances the system's adaptability to environmental changes.

[0052] (5) The inner optimization strategy establishes a stiffness adjustment model based on the structural response characteristics of each fin ray, and dynamically adjusts the equivalent stiffness through a nested optimization strategy, so that the fin ray structure response is closer to the optimal dynamic deformation state, thereby improving the structural compliance.

[0053] (6) The present invention introduces a trajectory tracking controller based on the model predictive control method, which enables the drive mechanism to achieve fast, smooth, and precise adjustment and tracking response without overshoot, thereby building a complete perception-decision-control closed-loop path.

[0054] (7) According to the method of the present invention, the method realizes efficient coordinated fluctuation of the fin surface in time and space by precisely controlling the spanwise fluctuation, chordwise phase and inward and outward movement of the fin root, thereby significantly improving the propulsion performance and attitude stability of the multi-ray driven pectoral fin, so that it has the ability to operate stably. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of a method for controlling spanwise and chordwise coordinated fluctuations of a fin-ray driven bionic pectoral fin according to an embodiment of the present invention.

[0056] Figure 2This is a three-dimensional structural diagram of the fin-driven bionic pectoral fin in an example of the present invention.

[0057] Figure 3 This is a diagram of the control signal for the fin-ray driven bionic pectoral fin root fluctuation angular displacement in an example of the present invention.

[0058] Figure 4 This is a diagram of the control signal for the fin-ray driven bionic pectoral fin root deflection angular displacement in an example of the present invention.

[0059] Figure 5 This is a schematic diagram of the smooth control of the fin root fluctuation of the bionic pectoral fin rhythm signal generator in an example of the present invention.

[0060] Figure 6 This is a schematic diagram of the smooth control of the fin root deflection of the bionic pectoral fin rhythm signal generator in an example of the present invention.

[0061] Figure 7 This is a schematic diagram of sensor arrangement and acquisition in the bionic pectoral fin sensing system in an example of the present invention.

[0062] Figure 8 It is a flow chart of the outer layer optimization strategy during the spanwise and chordwise coordinated fluctuation of the fin-ray driven bionic pectoral fin in an example of the present invention.

[0063] Figure 9 It is a flow chart of the inner layer optimization strategy during the spanwise and chordwise coordinated fluctuation of the fin-ray driven bionic pectoral fin in an example of the present invention.

[0064] Figure 10 This is a flowchart of the DDPG network training process during the spanwise and chordwise coordinated fluctuations of the fin-ray driven bionic pectoral fin in an example of the present invention.

[0065] Figure 11 It is a relationship diagram of the inner-outer layer optimization strategy of the span-wise and chord-wise coordinated fluctuations of the bionic pectoral fin in an example of the present invention.

[0066] Figure 12 This is a control block diagram of the bionic pectoral fin trajectory tracking controller in an example of the present invention.

[0067] Figure 13 It is a structural schematic diagram of the span-wise and chord-wise coordinated wave control device of the pectoral fin rays in an example of the present invention. DETAILED DESCRIPTION

[0068] In order to make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0069] Please refer to the attached Figures 1 to 13 The embodiment of the present invention provides a method, device and medium for controlling the spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin. The method comprises the following steps: (1) based on the skeletal structure of the pectoral fin of fish, a fin-ray driven bionic pectoral fin consisting of independently driven fin roots, elastic fin rays and fin surfaces is designed, the spanwise fluctuation of the fin surface is formed by the active fluctuation of the fin roots and the passive fluctuation of the fin rays, the fluctuation phase difference between the fin roots forms the chordwise fluctuation of the fin surface, and the deflection phase difference between the fin roots forms the adduction and abduction motion of the fin surface; at the same time, a fin root kinematic model is constructed based on the pectoral fin gait characteristic data, and the definition is the fin root motion control parameter; (2) in order to make the fin root kinematic model continuously output a smooth, continuous and rhythmic control signal, a rhythmic signal generator of an improved central pattern generator is constructed with the Hopf oscillator as the core; the rhythmic signal generator is matched with the input parameters of the fin root kinematic model, and the fin root motion control parameter is changed by changing the rhythmic signal generator. The rhythmic signal generator processes the output of the fin root kinematic model and outputs a smooth fin root angular displacement to achieve the three-dimensional coordinated movement of the fin-driven bionic pectoral fin in the span direction, chord direction, inward and outward directions. The smooth fin root angular displacement is defined as (3) During the movement of the pectoral fins, the pectoral fins are mounted on a six-dimensional force sensor to collect the pectoral fin thrust and torque in real time, and the local fluid load is obtained using a strain gauge array. , use the angle encoder to obtain the actual deflection angular displacement of the fin root during movement , using bending sensors to collect the actual angular displacement of the fin root during movement and the passive fin wave angular displacement , the actual deflection angular displacement of the fin root , the actual angular displacement of the fin root and the passive angular displacement of the fin ray The linear displacement and equivalent stiffness of the fin are calculated , the actual angular displacement of the fin root is defined as (4) Based on the deep deterministic policy gradient algorithm, the inner optimization strategy and outer optimization strategy of reinforcement learning are constructed; the outer optimization strategy adopts the weighted combination of maximizing propulsion performance and maximizing posture stability as the optimization goal, and optimizes the optimal fin root motion control parameters. , achieving a balance between propulsion performance and attitude stability, and enhancing the system's adaptability to environmental changes; the inner optimization strategy establishes a stiffness adjustment model based on the structural response characteristics of each fin, and dynamically adjusts the equivalent stiffness through a nested optimization strategy , so that the fin structure response is closer to the optimal dynamic deformation state, thereby improving the structural compliance; (5) In order to ensure accurate tracking of the expected angular displacement of the fin root generated by the outer layer optimization strategy , a trajectory tracking controller based on the model predictive control method is introduced. The trajectory tracking controller is constructed on the basis of the outer control trajectory and the inner stiffness adjustment. Based on the trajectory prediction and rolling optimization mechanism, the actual angular displacement of the current fin root is received. Desired angular displacement with the fin root The error between them is used to predict the future state evolution and to obtain the optimal control input by rolling solution. , thereby achieving high-precision, low overshoot and strong robustness trajectory tracking execution.

[0070] Specific embodiments of the present invention are further described below.

[0071] See Figure 1 Based on the structural characteristics of fish pectoral fins, this method and device designs a fin-ray-driven biomimetic pectoral fin mechanism. On this basis, a three-dimensional coordinated control system is constructed that integrates structural drive, rhythm generation, state perception, control optimization, and execution feedback. The system consists of three independently driven fin rays, a CPG rhythm signal generator, a state perception module, a two-layer DDPG reinforcement learning optimization controller, and an MPC execution controller. These functional modules form a complete perception-decision-execution closed-loop path, enabling coordinated control and efficient actuation of the biomimetic pectoral fin in complex hydrodynamic environments.

[0072] S1. See Figure 2 A fin-ray driven bionic pectoral fin structure provided in an embodiment of the present invention includes a rigid fin bone 1, a rigid fin root 2, an elastic fin ray 3 and a flexible fin membrane 4; the rigid fin bone 1 and the rigid fin root 2 are made of polylactic acid (PLA) material and are integrally formed through a three-dimensional printing process, which play a supporting and motion connection role in the pectoral fin structure; the fin ray 3 is made of a superelastic shape memory alloy plate, which has a large deformation ability and excellent structural flexibility, and is used to drive the pectoral fin to perform multi-degree-of-freedom wave motion; the fin membrane 4 is made of a hydrogel material with good flexibility and is arranged between the fin rays to maintain the continuity of the fin surface and enhance the hydrodynamic coupling effect.

[0073] Specifically, the first fin root 201 can only perform spanwise undulation motion, while the second and third fin roots 202 and 203 can further perform inward and outward motion on the basis of spanwise undulation. The active undulation of the fin root drives the passive undulation of the fin rays to form spanwise undulation of the flexible fin membrane 4. The undulation phase difference of each fin root 3 constitutes the chordwise undulation on the flexible fin membrane 4, and the deflection phase difference of each fin root 3 realizes the coordinated inward and outward motion of the fin surface, thereby realizing three-dimensional dynamic propulsion of the pectoral fin in space.

[0074] Specifically, since the fin ray 3 has a thin plate structure, when the fin root 2 generates an undulating motion, the fin ray 3 can bend significantly in the normal direction; while when the fin root 2 performs a deflection motion, the fin ray 3 mainly rotates in the tangential direction, and the structural rigidity is relatively large, and almost no bending deformation occurs;

[0075] Specifically, in order to realize the spanwise, chordwise, adduction and abduction motion control of the fin-driven bionic pectoral fin, a fin root kinematic model is constructed by combining the gait characteristics of the fish pectoral fin motion and the bionic law; first, the fin root fluctuation angular displacement The following piecewise cosine function is used for modeling:

[0076] ;

[0077] Where, For Moment Root fin root fluctuation angular displacement, For the The amplitude of the root fin root angular displacement, For the The offset of the root fin root fluctuation angular displacement, It is The root fin root fluctuation motion period ratio divides the motion period into four parts. , is the period of the wave motion, For time;

[0078] For details, see Figure 3 ,when , , , , , , , , , , , , , , , , When , the angular displacement of the 1st, 2nd and 3rd fin roots fluctuates.

[0079] Secondly, the fin root deflection angle displacement The modeling is as follows:

[0080] ;

[0081] Where, For Moment Root fin root deflection angular displacement, For the The amplitude of the angular displacement of the root fin, For the Offset of the root fin deflection angular displacement, It is The root fin root deflection motion period ratio, which divides the period into two parts, , is the deflection motion period, For time;

[0082] For details, see Figure 4 ,when , , , , , , , When , the 1st, 2nd and 3rd fin roots have deflection angular displacements.

[0083] Furthermore, the fin root motion control parameters , The values are 1, 2, and 3. The elastic fin ray generates fluctuations under the active drive of the fin root by relying on the elastic properties of its own material, and drives the flexible fin surface to achieve passive spanwise fluctuations.

[0084] S2. In order to make the fin root kinematic model continuously output smooth, continuous, rhythmic and highly bio-simulated control signals during execution, an improved central pattern generator based on the Hopf oscillator is further introduced as a rhythmic signal generator to dynamically modulate and smooth the control trajectory in real time.

[0085] Specifically, the rhythmic signal generator, based on the Hopf-type oscillation equation, constructs a dual-channel oscillator structure consisting of the fin root fluctuation angular displacement and the fin root deflection motion angular displacement. The core goal of this structure is to dynamically track the fin root control signal and perform second-order filtering to avoid execution discontinuities caused by sudden angle changes, while enhancing the control command's derivability, smoothness, and actuator response stability.

[0086] Specifically, the CPGs module is embedded into the control path of the fin root kinematic model, so that the input of the rhythm signal generator is the fin root angular displacement. and , defined as follows:

[0087] ;

[0088] Where, After smoothing by the rhythm signal generator Output, It's about The output of the rhythm signal generator, After smoothing by the rhythm signal generator Output, It's about The output of the rhythm signal generator, is the response frequency (the larger the faster the response), is the damping ratio (controls the smoothness), is time, the smoothed fin root angular displacement is , The value is 1, 2, 3;

[0089] Specifically, by changing the fin motion control parameters , the control signal output by the rhythm signal generator Control signals with fin ray kinematic model Similarly, the three-dimensional coordinated movement of the fin-driven bionic pectoral fin in the span direction, chord direction, inward and outward directions is achieved;

[0090] Specifically, to evaluate the effectiveness of the rhythm signal generator as a smoother in periodic signal generation, see Figure 5 , when the fin root fluctuation parameter in the first cycle is set to , , , , , , , , , , , , , , , , , and in the second period the fin root fluctuation parameter changes to , , , , , , , , , , , , , , , , , , From the local enlarged image, we can see that the control signal after being processed by the rhythm signal generator than the original control signal The output of is smoother; see Figure 6 , when the fin root deflection parameter in the first cycle is set to , , , , , , , , and in the second cycle the fin root deflection parameter changes to , , , , , , , From the local enlarged image, we can see that the control signal after being processed by the rhythm signal generator than the original control signal The output is smoother.

[0091] Furthermore, this module is located between the trajectory generation layer and the execution layer in the actual control path, and can be seamlessly coupled with the reinforcement learning strategy and state perception module, by outputting a three-channel smooth signal with rhythmic characteristics ( )and( ) are used to drive the synchronous control of the three fin roots in the span, chord, adduction and abduction directions, respectively, to construct a multi-fin ray collaborative propulsion mechanism with physical continuity and neural rhythm characteristics.

[0092] Furthermore, the above-mentioned rhythmic signal generator not only has a good approximation capability to the input signal of the fin kinematic model, but also can significantly improve the rhythmicity, smoothness and dynamic differentiability of the output angle trajectory. It is particularly suitable for processing non-continuous or edge-changing control signals such as piecewise defined functions, trapezoidal waves, and cosine superposition.

[0093] S3. During the movement of the pectoral fins, an integrated multi-source sensor perception system was constructed to achieve comprehensive perception and accurate observation of the propulsion state. The system rationally arranges sensors at key structural parts to obtain multi-dimensional physical quantity data of the pectoral fins in real time during the movement, thereby providing the controller with high-quality input required for state estimation. The pectoral fin thrust and torque are collected in real time by the six-dimensional force sensor, and the local fluid load is obtained by using the strain gauge array. , use the angle encoder to obtain the actual deflection angular displacement of the fin root during movement , using bending sensors to collect the actual angular displacement of the fin root during movement and the passive angular displacement of the fin ray , the actual deflection angular displacement of the fin root , the actual angular displacement of the fin root and the passive angular displacement of the fin ray The linear displacement and equivalent stiffness of the fin are calculated ;

[0094] For details, see Figure 7 ,The sensing system consists of a six-dimensional force sensor, a strain gauge array, an angle encoder and a bending sensor.

[0095] Furthermore, the data collection and processing process of the perception system is as follows:

[0096] Furthermore, the bionic pectoral fin is installed under the six-dimensional force sensor to collect the combined thrust and combined torque on the pectoral fin surface in the three-axis direction during movement in real time. The measured combined thrust and torque vectors are:

[0097] ,in, They are The resultant thrust on the pectoral fin surface in the direction of Respectively around The resultant moment on the axial pectoral fin surface;

[0098] Furthermore, strain gauge arrays are arranged at the root of the fin, at the fin root, in the middle of the fin ray and at the end of the fin ray, and the distribution of the local fluid load is measured to be , Respectively Root of the root fin, Middle and lower root fin rays The fluid load at the end of the root fin is used as the basis to construct the fin fluid pressure function along the length of the fin using quadratic interpolation:

[0099] ;

[0100] Where, For the The length of the fin installed at the root of the root fin, is a constant, which is obtained from the fluid load data at the fin root, the middle of the fin ray and the end of the fin ray;

[0101] Specifically, The fluid pressure at the end of the root fin is , No. The fluid moment at the end of the root fin is .

[0102] Furthermore, the actual angular displacement of the fin root is acquired by using an angle encoder. The actual angular displacement of the fin root is obtained by a flexible bending sensor. and the passive angular displacement of the fin ray , combining the geometric model and material response relationship, the linear displacement of the fin tip is calculated as:

[0103] ;

[0104] Where, for Moment The end displacement of the root ray normal, for Moment The tangential displacement of the root ray, For the The length of the fin ray installed at the root of the root fin;

[0105] Specifically, the equivalent stiffness The calculation formula is:

[0106] ;

[0107] Where, For the The equivalent stiffness of the root fin (the larger the root fin, the harder it is to bend). The first Passive wave angular displacement of root fin ray;

[0108] Specifically, in order to verify the accuracy of the fluid load inversion results based on strain measurement, the first Equivalent stiffness of root fin ray and compare it with the equivalent stiffness inversely deduced from the fluid moment If the two are close, it means that the estimated fluid load distribution has good physical consistency, where For the Elastic modulus of the root fin ray, For the Sectional moment of inertia of the root ray.

[0109] Furthermore, all sensor data are uniformly input into the state perception module, and after filtering, normalization, and time alignment, the structured state vector required by the reinforcement learning controller is formed:

[0110] ;

[0111] Where, ;

[0112] Specifically, the state vector is used as the environmental observation input of the outer optimization strategy to evaluate the system propulsion performance and attitude stability; at the same time, The data is also synchronously transmitted to the inner layer optimization strategy for dynamic correction of fin ray stiffness.

[0113] S4. In order to balance the relationship between propulsion performance, attitude stability and fin motion compliance, an optimization strategy is constructed based on the deep deterministic policy gradient algorithm. The optimization strategy design adopts an inner-outer nested two-layer reinforcement learning structure. The inner optimization strategy is used for local structural stiffness adjustment, and the outer optimization strategy is used for global motion parameter optimization. The two work together to form a dynamic adaptive control closed loop.

[0114] Further, see Figure 8 The outer optimization strategy adopts a weighted combination of maximizing propulsion performance and maximizing attitude stability as the outer optimization goal, and obtains the optimal fin motion control parameters through DDPG algorithm learning. (spanwise amplitude, spanwise frequency, spanwise phase, spanwise motion cycle ratio, adduction and abduction amplitude, adduction and abduction frequency, adduction and abduction phase, adduction and abduction motion cycle ratio), achieving a balance between propulsion performance and stability, and enhancing the system's adaptability to environmental changes;

[0115] Specifically, the equation of the outer optimization objective is:

[0116] ;

[0117] Where, is the weight coefficient, when When the propulsion performance is the main factor, When doing so, the main focus is on posture stability;

[0118] Specifically, the propulsion performance calculation formula is:

[0119] ;

[0120] Where, is the effective thrust generated per unit power consumption, which characterizes the propulsion performance of the pectoral fin; For power consumption, is the thrust in the x direction.

[0121] Specifically, the posture stability calculation formula is:

[0122] ;

[0123] Where, represents the attitude disturbance of the pectoral fin in the yaw degree of freedom, represents the attitude disturbance of the pectoral fin in the pitch degree of freedom, Represents the variance function, and uses variance to represent the degree of fluctuation of force and torque. The smaller it becomes, the better the posture stability.

[0124] Further, see Figure 9 The inner layer optimization strategy takes maximizing the structural compliance of the fin ray during motion as the inner layer optimization objective function while keeping the optimal fin ray motion control parameters unchanged, and adjusts the equivalent stiffness parameters of each fin ray. , the actual deformation trajectory under fluid load has good structural compliance;

[0125] Specifically, the equation of the inner optimization objective is:

[0126] ;

[0127] Where, The equivalent stiffness is Time Passive angular displacement of the root fin ray, The equivalent stiffness is Time The end displacement of the root ray normal, is the weighting coefficient of the two parts, is the pectoral fin movement cycle, For time;

[0128] Specifically, the first term in the above formula measures the passive bending response of the fin under the action of fluid and controls its degree of flexible deformation. The second term measures the smoothness of the fin motion curve (limiting the stiffness adjustment from being too frequent or drastic to avoid material fatigue or uncontrollable deformation).

[0129] Further, see Figure 10 ,DDPG training process includes policy network and evaluation network, combined with ,experience replay and target network soft update mechanism to ensure ,training stability and convergence effect.

[0130] Further, see Figure 11 In the inner-outer nested two-layer reinforcement learning optimization strategy, the outer optimization strategy realizes high-level decision-making and control of the fin root motion behavior goals, and the inner optimization strategy ensures the flexibility of the fin ray dynamic response by adjusting the stiffness. The two-layer strategy works synergistically to effectively improve the environmental adaptability, posture stability and propulsion performance of the bionic pectoral fin control system under dynamic conditions.

[0131] S5. To ensure accurate tracking of the desired fin root angular displacement generated by the outer layer optimization strategy , the system introduces a trajectory tracking controller based on the model predictive control method at the execution layer; see Figure 12The trajectory tracking controller is constructed on the basis of outer control trajectory and inner stiffness adjustment, based on trajectory prediction and rolling optimization mechanism, using the desired angular displacement of the fin root and the actual angular displacement of the fin root Calculate the displacement error, build a dynamic evolution model within the prediction window, and solve the problem in real time by minimizing the sum of squares of the displacement error and the weighted sum of squares of the control inputs. Output a series of future control inputs and obtain the optimal control input in each control cycle. , thereby achieving high-precision, low overshoot and strong robustness trajectory tracking execution.

[0132] Specifically, the objective function of the trajectory tracking controller is:

[0133] ;

[0134] Where, yes Prediction step length Place Expected angular displacement of the root fin, , The prediction step length Place The actual angular displacement of the root fin, , The prediction step length The first The optimal control input of the root fin root is obtained in each control cycle , the superscript “ˆ” only represents the optimal value and does not change the meaning of the letter; is the error weight matrix, is the control cost matrix, is the prediction step length.

[0135] Specifically, the objective function of the trajectory tracking controller of the first fin root is solved as follows:

[0136] ;

[0137] Where, yes Prediction step length The expected angular displacement of the first fin root at , The prediction step length The actual angular displacement of the first fin root at , The prediction step length The optimal control input of the first fin root at the position is obtained in each control cycle. , is the error weight matrix, is the control cost matrix, is the prediction step size, A and B represent the linear system state space matrices, the superscript “min” represents the minimum value, and the superscript “max” represents the maximum value;

[0138] Specifically, the objective function of the trajectory tracking controller of the second fin root is solved as follows:

[0139] ;

[0140] Where, yes Prediction step length The expected angular displacement of the second fin root is , The prediction step length The actual angular displacement of the second fin root at , The prediction step length The optimal control input of the second fin root at the position is obtained in each control cycle. , is the error weight matrix, is the control cost matrix, is the prediction step size, A and B represent the linear system state space matrices, the superscript “min” represents the minimum value, and the superscript “max” represents the maximum value;

[0141] Specifically, the objective function of the trajectory tracking controller of the third fin root is solved as follows:

[0142] ;

[0143] Where, yes Prediction step length The expected angular displacement of the third fin root is , The prediction step length The actual angular displacement of the third fin root is , The prediction step length The optimal control input of the third fin root at each control cycle is obtained. , is the error weight matrix, is the control cost matrix, For the prediction step size, A and B represent the state space matrices of the linear system, the superscript “min” represents the minimum value, and the superscript “max” represents the maximum value.

[0144] Furthermore, the trajectory tracking controller is used to enable the drive mechanism to achieve fast, smooth, and overshoot-free target angle tracking response, thereby constructing a complete perception-decision-control closed-loop path.

[0145] The following describes a fin-ray driven bionic pectoral fin span-wise and chord-wise coordinated motion control device provided by the present invention. The underwater vehicle active stabilization control device described below and the underwater vehicle active stabilization control method described above can be referenced to each other.

[0146] For details, see Figure 13 The fin-ray driven bionic pectoral fin span-wise and chord-wise coordinated motion control device includes: a drive control module, a state perception module, a signal processing module and a feedback adjustment module.

[0147] Specifically, the driving control module is used to output the control signal of the rhythm signal generator and drive the actuator at the fin root to perform corresponding movement;

[0148] Specifically, the state perception module collects the state vector of the first moment of the pectoral fin movement through sensors. The state variables include the pectoral fin thrust and torque, the local fluid load, the actual fin root fluctuation angular displacement, the passive fin ray fluctuation angular displacement, and the actual fin root deflection angular displacement, which are called in real time by the feedback adjustment module.

[0149] Specifically, the signal processing module generates the fin root motion control parameters through calculation of the outer layer optimization strategy, the inner layer optimization strategy and the trajectory tracking controller;

[0150] Specifically, the feedback adjustment module compares the data results of the state perception module with the current control parameters, generates dynamic correction information, triggers the outer optimization strategy and the inner optimization strategy through the interrupt mechanism to re-update the action strategy, and at the same time ensures that the fin root tracks the desired angular displacement of the fin root through the trajectory tracking controller.

[0151] On the other hand, the present invention also provides a motion control program instruction, which is stored in the computer-readable storage medium composed of a Raspberry Pi as a host and a field programmable logic gate array processing unit as a slave, and the computer-readable storage medium is used to execute the motion control program instruction to implement a fin-driven bionic pectoral fin span-wise and chord-wise collaborative motion control method provided by the above method.

[0152] Specifically, the aforementioned motion control program instructions may exist in the form of source code, bytecode, intermediate representation, or platform executable file;

[0153] Specifically, during the execution of the motion control program instructions, the control logic and data flow are as follows: (1) The current propulsion mission objectives and flow field information are input by the external task management system; (2) The outer DDPG strategy network generates motion parameters based on the current state vector; (3) The inner stiffness optimization module optimizes the fin stiffness based on the reference trajectory and feedback displacement. ; (3) The MPC module predicts future control inputs, and the instruction drive module outputs execution signals; (4) The data acquisition module synchronously collects sensor information and transmits it back for the next cycle control iteration.

[0154] On the other hand, the present invention also provides a computer-readable storage medium, which is composed of a master-slave control architecture, and its control program is deployed in an integrated computing platform, the host adopts a Raspberry Pi microcomputer, and the slave adopts a field programmable logic gate array processing unit; motion control program instructions are stored thereon, and the motion control program instructions are executed by the controller to implement a fin-driven bionic pectoral fin span-wise and chord-wise collaborative motion control method provided by the above method.

[0155] It should be noted that the device structure and module division described in the above specification are only schematic descriptions of the embodiments of the present invention and do not constitute a limitation on the scope of protection of the present invention. In the specific implementation process, the functional modules or units described can be physically integrated according to actual needs, or they can be logically independent of each other. For example, as independently described modules, in actual deployment, they can be physically separated independent units, or they can be integrated into the same controller or computing platform; the communication connection between modules can be achieved in various forms such as signal lines, data buses, and wireless transmission. Those skilled in the art can make reasonable adjustments and implementations based on the technical solution of the present invention and its description in combination with specific application scenarios without paying creative labor.

[0156] The background section of the present invention may contain background information about the problem or environment of the present invention, but does not necessarily describe the prior art. Therefore, the inclusion of content in the background section is not an admission by the applicant that the prior art is present.

[0157] The above description is provided as a further detailed description of the present invention in conjunction with specific embodiments, and the specific implementation of the present invention is not limited to these descriptions. Obviously, those skilled in the art may make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if such modifications and variations fall within the scope of the claims and their equivalents, the present invention is intended to include such modifications and variations.

Claims

1. A fin-ray driven bionic pectoral fin span-wise and chord-wise coordinated motion control method, characterized in that: The following steps are involved: S1. Based on the skeletal structure of the pectoral fin of fish, a fin-ray driven bionic pectoral fin consisting of independently driven fin roots, elastic fin rays and fin surfaces is designed. The spanwise fluctuation of the fin surface is formed by the active fluctuation of the fin roots and the passive fluctuation of the fin rays. The chordwise fluctuation of the fin surface is formed by the fluctuation phase difference between the fin roots. The inward and outward movement of the fin surface is formed by the deflection phase difference between the fin roots. At the same time, according to the characteristic data of the pectoral fin gait, a fin root kinematic model is constructed to define is the fin root motion control parameter; S2. In order to make the fin root kinematic model continuously output a smooth, continuous and rhythmic control signal, a rhythmic signal generator of an improved central pattern generator is constructed with the Hopf oscillator as the core; the rhythmic signal generator is matched with the input parameters of the fin root kinematic model, and the fin root motion control parameters are changed by changing the rhythmic signal generator. The rhythmic signal generator processes the output of the fin root kinematic model and outputs a smooth fin root angular displacement to achieve the three-dimensional coordinated movement of the fin-driven bionic pectoral fin in the span direction, chord direction, inward and outward directions. The smooth fin root angular displacement is defined as ; S3. During the movement of the pectoral fins, the pectoral fins are mounted on a six-dimensional force sensor to collect the pectoral fin thrust and torque in real time, and the local fluid load is obtained using a strain gauge array. , use the angle encoder to obtain the actual deflection angular displacement of the fin root during movement , using bending sensors to collect the actual angular displacement of the fin root during movement and the passive fin wave angular displacement , the actual deflection angular displacement of the fin root , the actual angular displacement of the fin root and the passive angular displacement of the fin ray The linear displacement and equivalent stiffness of the fin are calculated , the actual angular displacement of the fin root is defined as ; S4. Construct an inner optimization strategy and an outer optimization strategy for reinforcement learning based on a deep deterministic policy gradient algorithm. The outer optimization strategy uses a weighted combination of maximizing propulsion performance and maximizing posture stability as the optimization goal to optimize the optimal fin root motion control parameters. The inner optimization strategy establishes a stiffness adjustment model based on the structural response characteristics of each fin, and dynamically adjusts the equivalent stiffness through a nested optimization strategy. ; S5. To ensure accurate tracking of the desired fin root angular displacement generated by the outer layer optimization strategy , a trajectory tracking controller based on the model predictive control method is introduced. The trajectory tracking controller is constructed on the basis of the outer control trajectory and the inner stiffness adjustment. Based on the trajectory prediction and rolling optimization mechanism, the actual angular displacement of the current fin root is received. Desired angular displacement with the fin root The error between them is used to predict the future state evolution and to obtain the optimal control input by rolling solution. , thereby achieving high-precision, low overshoot and strong robustness trajectory tracking execution.

2. The method for controlling spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin according to claim 1, characterized in that: The fin-ray driven bionic pectoral fin is composed of an independently driven fin root, a passively deformed elastic fin ray, and a passively deformed flexible fin surface. The first fin root realizes spanwise fluctuation, and the second and third fin roots further have the coupled motion capability of adduction and abduction on the basis of spanwise fluctuation. The active fluctuation of the fin root and the passive fluctuation of the fin ray realize spanwise fluctuation of the entire fin surface, the fluctuation phase difference realizes chordwise fluctuation of the entire fin surface, and the deflection phase difference realizes adduction and abduction motion of the entire fin surface. The equation of the fin root kinematic model is: ； ； Where, For Moment Root fin root fluctuation angular displacement, For the The amplitude of the root fin root angular displacement, For the The offset of the root fin root fluctuation angular displacement, It is The root fin root wave motion period ratio, the period is divided into four parts, , is the period of wave motion; For Moment Root fin root deflection angular displacement, For the The amplitude of the angular displacement of the root fin, For the Offset of the root fin deflection angular displacement, It is The root fin root deflection motion period ratio, which divides the period into two parts, , is the deflection motion period, is time, the fin root motion control parameter , When the value is 1, 2, or 3, the elastic fin ray generates fluctuations under the active drive of the fin root by relying on the elastic properties of its own material, and drives the flexible fin surface to achieve passive spanwise fluctuations.

3. The method for controlling spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin according to claim 1, characterized in that: The rhythmic signal generator is constructed by the improved central pattern generator and the fin root kinematic model. The equation of the rhythmic signal generator is as follows: ； Where, After smoothing by the rhythm signal generator Output, It's about The output of the rhythm signal generator, After smoothing by the rhythm signal generator Output, It's about The output of the CPG, Is the response frequency. The larger the value, the faster the response. is the damping ratio, which controls the smoothness, is time, the smoothed fin root angular displacement is The rhythmic signal generator not only has a good approximation capability to the original control signal, but also has the capability to maintain the rhythmicity, smoothness and conductibility of the trajectory.

4. The method for controlling spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin according to claim 1, characterized in that: The collection and processing procedures of the pectoral fin thrust and torque, the local fluid load, the actual fin root fluctuation angular displacement, the passive fin ray fluctuation angular displacement, the actual fin root deflection angular displacement and the fin ray motion line displacement are as follows: First, the pectoral fin thrust and torque on the pectoral fin surface in the three-axis direction are collected in real time by a six-dimensional force sensor installed above the pectoral fin base, and the pectoral fin thrust and torque are recorded as , They are Directional thrust of the pectoral fin surface, Respectively around The moment of the pectoral fin surface is measured; at the same time, strain gauge arrays are arranged at the fin root, the middle of the fin ray and the end of the fin ray, and the distribution of the local fluid load is measured as follows: , Respectively Root of the root fin, Middle and lower root fin rays The fluid load at the end of the root fin is constructed by quadratic interpolation to obtain the fin fluid pressure function along the length of the fin: ； Where, For the The length of the fin installed at the root of the root fin, is a constant, which is obtained by the fluid load data at the fin root, the middle of the fin ray and the end of the fin ray. The fluid pressure at the end of the root fin is , No. The fluid moment at the end of the root fin is ; Secondly, the actual deflection angular displacement of the fin root is collected using an angle encoder The actual angular displacement of the fin root is obtained by a flexible bending sensor. and the passive angular displacement of the fin ray , combining the geometric model and material response relationship, the linear displacement of the fin tip is calculated as: ； Where, for Moment The end displacement of the root ray normal, for Moment The tangential displacement of the root ray, For the Root ray length; equivalent stiffness The calculation formula is: ； Where, For the The equivalent stiffness of the root fin, the larger the value, the harder it is to bend. The first Passive oscillation angular displacement of the root fin ray, The values are 1, 2, 3; Finally, the above data are uniformly sampled, filtered, normalized and feature fused by the state perception module to form a state vector: ； Where, , the state vector is used as the environmental observation input of the outer optimization strategy to evaluate the system propulsion performance and attitude stability; at the same time, The data is also synchronously transmitted to the inner layer optimization strategy for dynamic correction of fin ray stiffness.

5. The method for controlling spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin according to claim 1, characterized in that: The outer optimization strategy takes the maximization of propulsion performance and posture stability as the outer optimization goals, and obtains the optimal fin root motion control parameters through deep deterministic policy gradient algorithm learning. , the superscript "˜" only represents the optimal value of the motion control parameter and does not change the meaning of the letter. The equation of the outer optimization objective is: ； Where, is the weight coefficient, The calculation equation is: ； Where, is the effective thrust generated per unit power consumption, For power consumption, represents the attitude disturbance of the pectoral fin in the yaw degree of freedom, represents the attitude disturbance of the pectoral fin in the pitch degree of freedom, represents the variance function, when The smaller the value, the better the attitude stability; the inner optimization strategy takes maximizing the structural flexibility during the fin root movement as the inner optimization goal while keeping the optimal fin root motion control parameters unchanged, and adjusts the equivalent stiffness of each fin ray , the actual deformation trajectory under fluid load has good structural compliance; the equation of the inner layer optimization objective is: ； The first term measures the passive bending response of the fin under the action of the fluid, controlling its degree of compliance. The second term measures the smoothness of the fin motion curve, limiting the frequent or drastic adjustment of stiffness to avoid material fatigue or uncontrollable deformation. The equivalent stiffness is Time Passive oscillation angular displacement of the root fin ray, The equivalent stiffness is Time The end displacement of the root ray normal, is the weighting coefficient of the two parts, is the pectoral fin movement cycle, The outer optimization strategy realizes high-level decision-making and control of the fin root motion behavior target, and the inner optimization strategy ensures the flexibility of the fin ray dynamic response by adjusting the stiffness. The two-layer strategy works together to effectively improve the environmental adaptability, posture stability and propulsion performance of the bionic pectoral fin control system under dynamic conditions.

6. The method for controlling spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin according to claim 1, characterized in that: The trajectory tracking controller uses a model predictive control approach to ensure accurate tracking of the desired fin root angular displacement generated by the outer layer optimization strategy. Based on trajectory prediction and rolling optimization mechanism, the expected angular displacement of the fin root is used and the actual angular displacement of the fin root Calculate the displacement error, build a dynamic evolution model within the prediction window, and solve the problem in real time by minimizing the sum of squares of the displacement error and the weighted sum of squares of the control inputs. Output a series of future control inputs and obtain the optimal control input in each control cycle. , in order to achieve accurate tracking of the target trajectory; the equation of the trajectory tracking controller is: ； Where, yes Prediction step length Place Expected angular displacement of the root fin, , The prediction step length Place The actual angular displacement of the root fin, , The prediction step length The first The optimal control input of the root fin root is obtained in each control cycle , the superscript "ˆ" only represents the optimal value of the control input and does not change the meaning of the letter; is the error weight matrix, is the control cost matrix, The step length is predicted; the trajectory tracking controller is used to enable the drive mechanism to achieve fast, smooth, and overshoot-free target angle tracking response, thereby constructing a complete perception-decision-control closed-loop path.

7. The method for controlling spanwise and chordwise coordinated motion of a fin-ray driven bionic pectoral fin according to claim 1, characterized in that: The fin-ray driven bionic pectoral fin is made of rigid fin bones, rigid fin roots, elastic fin rays and flexible fin membranes; the fin bones and fin roots are made of materials with good structural strength and molding precision, the fin rays are made of materials with high elasticity and lightweight properties, and the fin membrane is made of flexible and deformable materials.

8. A fin-ray driven bionic pectoral fin span-wise and chord-wise coordinated motion control device, characterized in that: include: A drive control module, configured to output a control signal to the rhythmic signal generator and drive the actuator at the fin root to perform corresponding motion; A state perception module collects the state vector of the first moment of the pectoral fin movement through sensors. The state variables include the pectoral fin thrust and torque, the local fluid load, the actual fin root fluctuation angular displacement, the passive fin ray fluctuation angular displacement, and the actual fin root deflection angular displacement, which are called in real time by the feedback adjustment module. A signal processing module generates the fin root motion control parameters through calculation of an outer layer optimization strategy, an inner layer optimization strategy, and a trajectory tracking controller; The feedback adjustment module compares the data results of the state perception module with the current control parameters, generates dynamic correction information, triggers the outer optimization strategy and the inner optimization strategy through the interrupt mechanism to re-update the action strategy, and ensures that the fin root tracks the desired angular displacement of the fin root through the trajectory tracking controller.

9. A computer-readable storage medium containing motion control program instructions, characterized in that: The motion control program instructions are stored in the computer-readable storage medium composed of a Raspberry Pi as a host and a field programmable logic gate array processing unit as a slave. The computer-readable storage medium is used to execute the motion control program instructions to implement a fin-driven bionic pectoral fin span-wise and chord-wise coordinated motion control method as described in any one of claims 1 to 7.

Images