Multi-dimensional vibration control method, system and device based on model predictive control

Abstract

The application discloses a kind of multi-dimensional vibration control method, system and device based on model predictive control, method includes: to the kinematics and dynamics analysis of multi-dimensional vibration control device, establish linear dynamics equation near equilibrium point, establish prediction model according to linear dynamics equation;According to the output of prediction time domain by the control amount of control time domain according to prediction model estimation, according to the control target setting optimization target that output of prediction time domain approaches expected trajectory, set constraint condition, determine the optimization problem of control amount;According to the pose information of carrying equipment at t time determines expected trajectory;Optimization problem is solved, determine the optimal control amount sequence, and the first value in sequence is as the actual control amount of current time, control multi-dimensional vibration control device.The application can realize early detection, predict future, early control, reduce system delay effect, optimal control under complex constraint environment and multi-dimensional high-precision damping effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of vibration reduction technology, in particular to a multi-dimensional vibration reduction control method, system and device based on model predictive control. BACKGROUND

[0002] In the working scene of mobile carrying equipment such as vehicles and ships, the driving process is often too bumpy to affect the stability of the mobile carrying equipment, for example, in the scene of road bumps during driving and resistance to sea wave bumps during ship driving, which causes people on the carrying equipment to be seasick or car sick, and goods to be damaged. To solve this problem, a damping device is generally used to automatically compensate for the inertial rotation and inertial displacement caused by the bumping, acceleration, braking and the like of the carrying equipment during driving, so as to keep the goods, personnel and the like on the carrying equipment in a stable state and reduce the risk in the transportation process. Most of the existing damping devices use passive damping to reduce vibration, such as hydraulic devices or spring devices, but these damping devices have slow response speed and the damping effect is not obvious under high-frequency vibration. Therefore, the applicant has been committed to developing a multi-dimensional active damping device for vehicles or ships. Compared with passive and semi-active damping devices, the multi-dimensional active damping device can achieve better damping effect. The current multi-dimensional active damping device mainly uses traditional control theory, such as PID, optimal control, robust control and adaptive control.

[0003] Model predictive control is a new type of control method that has emerged in recent years. Compared with traditional control theory, this method can predict the future state of the system and actively handle various constraint conditions to achieve optimal control under multiple constraint conditions containing prediction information, which can achieve better control effect. It has been widely used in the field of autonomous driving cars, but its research and application in the field of multi-dimensional active damping devices are still very few. SUMMARY

[0004] To solve the above technical problems, the present application provides a multi-dimensional vibration reduction control method, system and device based on model predictive control, which fully considers various constraint conditions existing in the actual engineering application of the multi-dimensional active damping device, and realizes optimal control under multiple constraint conditions containing prediction information.

[0005] Specifically, the technical solutions of the present application are as follows:

[0006] The present application provides a multi-dimensional vibration reduction control method based on model predictive control, which is applied to a multi-dimensional vibration reduction device. The method comprises the following steps:

[0007] S100: kinematics and dynamics analysis is performed on the multi-dimensional vibration reduction device, a linear dynamics equation near the equilibrium point is established, and a prediction model is established according to the linear dynamics equation;

[0008] S200: estimating the output quantity in the prediction time domain according to the control quantity in the control time domain according to the prediction model, setting the optimization target according to the control target that the output quantity in the prediction time domain approaches the expected trajectory, setting the constraint condition, and determining the optimization problem of the control quantity;

[0009] S300: determining the expected trajectory according to the pose information of the carrying device at the t time;

[0010] S400: solving the optimization problem, determining the optimal control quantity sequence, taking the first value in the sequence as the actual control quantity at the current time, and controlling the multi-dimensional vibration damping device;

[0011] S500: t=t+1, returning to S300.

[0012] In some embodiments, the linear dynamic equation near the equilibrium point in S100 is:

[0013]

[0014] wherein, is the task space coordinate, is the task space velocity, M0 is the task space mass matrix at the equilibrium point, C0 is the task space damping matrix at the equilibrium point, K0 is the task space stiffness matrix at the equilibrium point, J0 is the Jacobian matrix at the equilibrium point, and u is the control force.

[0015] In some embodiments, the prediction model established according to the linear dynamic equation in S100 is

[0016] Y p (t+1|t)=S x Δx(t)+S y y(t)+S u ΔU m (t)

[0017] wherein, Y p (t+1|t) is the output sequence vector predicted at the t time for the future N times, ΔU m (t) is the input sequence vector predicted at the t time for the future m times, S x , S y , and S u are parameter matrices.

[0018] In some embodiments, S200 includes: estimating the output quantity in the prediction time domain according to the control quantity in the control time domain according to the prediction model, setting the optimization target according to the control target that the output quantity in the prediction time domain approaches the expected trajectory.

[0019] S201: determining the optimization target as

[0020] J(x(t), ΔU m (t)) = ||Γ y [Y p (t+1|t) - R(t+1)]|| 2 +||Γ u ΔU m (t)| 2

[0021] wherein R(t+1) is a desired trajectory point vector; Γ y is an output weight matrix; Γ u is an input weight matrix;

[0022] S202: obtaining an equivalent target value of the optimization target according to the prediction model

[0023]

[0024] wherein

[0025]

[0026]

[0027] In some embodiments, the constraint condition of S200 includes but is not limited to travel limit, motion space limit, actuator power / output limit.

[0028] In some embodiments, the optimization problem determined by S200 is

[0029]

[0030] s.t.LΔU m (t)≤b

[0031] In some embodiments, S300 includes determining the desired trajectory according to the pose information of the carrier device at time t.

[0032] The pose information of the carrier device at time t is measured by an inertial navigation unit, and the position and velocity of the corresponding dimension of the vibration reduction device task space are taken as input pose quantities r In , and the desired trajectory is set as r(t) = -r In .

[0033] In some embodiments, after measuring the pose information of the carrier device at time t by the inertial navigation unit, the obtained pose information is further filtered by a multi-sensor perception fusion algorithm.

[0034] The application also provides a multi-dimensional vibration reduction control system based on model prediction control, which is applied to a multi-dimensional vibration reduction device, and the system comprises:

[0035] The model establishing module is configured to perform kinematics and dynamics analysis on the multi-dimensional damping device, establish a linear dynamics equation near the equilibrium point, and establish a prediction model according to the linear dynamics equation.

[0036] The optimization module is configured to estimate an output quantity in a prediction time domain by controlling a control quantity in a control time domain according to the prediction model, set an optimization target according to a control target that the output quantity in the prediction time domain approaches the expected trajectory, set a constraint condition, and determine an optimization problem of the control quantity.

[0037] The measurement module is configured to determine the expected trajectory according to the pose information of the carrying device at the time t.

[0038] The control module is configured to solve the optimization problem, determine an optimal control sequence, take a first value in the sequence as an actual control quantity at the current time, and control the multi-dimensional damping device.

[0039] The application further provides a multi-dimensional damping device based on model prediction, comprising:

[0040] A base is fixed on the carrying device.

[0041] An inertial navigation unit is installed on the base and configured to measure the pose information of the carrying device.

[0042] A robot mechanism comprises a control unit, an execution unit and an encoder, wherein

[0043] The control unit is electrically connected with the inertial navigation unit, and the control unit comprises a memory and a processor, the memory is configured to store a control program, and the processor is configured to load and execute the control program to realize the multi-dimensional damping control method based on model prediction control.

[0044] The execution unit is electrically connected with the control unit and configured to perform operations according to the control instruction of the control unit.

[0045] The encoder is electrically connected with the control unit and the execution unit, configured to detect the actual execution of the execution unit and feed back to the control unit.

[0046] An upper platform is connected with the base through the robot mechanism and configured to carry people or objects.

[0047] Compared with the prior art, the application has at least one beneficial effect:

[0048] 1. Prediction of future state of the system: the application adopts model prediction control, predicts the state of the system at N future time points in advance at each time point, compares with the expected future trajectory, and performs advanced control, which has good foresight.

[0049] 2. Realize optimal control under complex constraint condition: The present application fully considers various constraints existing in actual engineering of the multi-dimensional damping device, including but not limited to stroke limit, movement space limit, actuator power / output limit, etc., and can realize optimal control without violating the above constraints, having good engineering practicability.

[0050] 3. Establish prediction model according to linearization equation of task space near equilibrium point, high calculation efficiency: The present application analyzes the working characteristics of the multi-dimensional active damping device, determines that it moves near the equilibrium point, establishes linearization dynamics equation of task space at the equilibrium point, and establishes prediction model according to the same, the obtained model form is simple and clear, the calculation efficiency is high, and the real-time performance of model prediction control is ensured.

[0051] 4. Reduce the influence of system delay: The model prediction control of the present application fully utilizes future information, and makes early prediction on future state of the system, realizes early control, can effectively reduce the influence of system delay, and further improves the damping effect of the damping device. BRIEF DESCRIPTION OF DRAWINGS

[0052] The above characteristics, technical features, advantages and implementation ways of the multi-dimensional damping control method, system and device based on model prediction control will be further described in the following in a clear and understandable manner combined with the drawings.

[0053] Figure 1 is a schematic diagram of model prediction control of the present application;

[0054] Figure 2 is a flow chart of a multi-dimensional damping control method based on model prediction control of the present application;

[0055] Figure 3 is a structural schematic diagram of a multi-dimensional damping device of the present application;

[0056] Figure 4 is a structural schematic diagram of a robot configuration which can be used in the embodiment of the present application;

[0057] Figure 5 is a mechanical structure diagram of the multi-dimensional damping device of the embodiment of the present application;

[0058] Figure 6 is a structural schematic diagram of the multi-dimensional damping device of the embodiment of the present application;

[0059] Figure 7 is the electric cylinder control force of each branch chain in the embodiment of the present application;

[0060] Figure 8 is the state quantity of the multi-dimensional damping device in the embodiment of the present application. DETAILED DESCRIPTION

[0061] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the specific embodiments of the present invention will be described below with reference to the accompanying drawings. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings and other embodiments can be obtained based on these drawings without inventive work.

[0062] To simplify the drawings, only portions relevant to the invention are schematically depicted in each figure; they do not represent the actual structure of the product. Furthermore, to simplify the drawings and facilitate understanding, in some figures, only one component with the same structure or function is schematically depicted or labeled. In this document, "one" not only means "only one" but also "more than one."

[0063] It should be further understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0064] It should be noted that, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be understood broadly. For example, they can refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediary; and internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in the present invention based on the specific circumstances.

[0065] In addition, in the description of the present application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0066] Reference Manual Figure 1 The core idea of ​​model predictive control is to perform predictive control. At time t, a prediction model is established based on the discrete state equation to estimate the output y(t+k|t), k=1,2,…,N for the next N moments. By optimizing the control quantity u(t+k|t), k=0,2,…,m-1 for the next m moments, the output quantity sequence y(t+k|t), k=1,2,…,N is made as close to the desired trajectory r(t) as possible, and the control quantity is made smooth. At the same time, various constraints on the input and output quantities are considered. Finally, the optimal control quantity sequence u(t+k|t), k=0,2,…,m-1 at time t is calculated, and the first value in the sequence is used as the actual control quantity at the current moment. The above process is repeated at each subsequent moment.

[0067] Model predictive control can achieve better control effect compared with traditional control theory through the method of predicting future, and it can actively deal with various constraints, which is very suitable for complex engineering environment. Because multi-dimensional active vibration reduction device is a standard multi-input multi-output system, and there are various constraints such as stroke limit, motion space limit, actuator power / output limit, etc., it is difficult to design an ideal control algorithm using traditional control theory, therefore the present application proposes to use model predictive control for design, which can realize optimal control under multi-constraints containing prediction information.

[0068] In one embodiment, reference is made to the description attached Figure 2 , the present application provides a kind of multi-dimensional vibration reduction control method based on model predictive control, it is applied to multi-dimensional vibration reduction device, described method includes the following steps:

[0069] Step one, the kinematics and dynamics analysis of multi-dimensional vibration reduction device is carried out, the linear dynamics equation near equilibrium point is established, and the prediction model is established according to the linear dynamics equation.

[0070] The dynamics equation of robot is established in the task space (also known as operation space, here it is the inertia space of the platform on the multi-dimensional active vibration reduction device)

[0071]

[0072] Among them, is the task space coordinate; is the mass matrix of robot in task space; is centrifugal force, coriolis force, friction, gravity, rigid force, damping force, etc. in task space;u is the control force applied by each actuator in robot joint space; is the velocity / force mapping jacobian matrix of robot, that is, it satisfies

[0073]

[0074] In the above formula, is the joint space velocity of robot; is the task space velocity of robot.

[0075] Because multi-dimensional active vibration reduction device generally works near the equilibrium point, linearization can be carried out near the equilibrium point, and the linear dynamics equation of small deviation is established, that is

[0076]

[0077] Wherein, M0 is the task space mass matrix at the equilibrium point; C0 is the task space damping matrix at the equilibrium point; K0 is the task space stiffness matrix at the equilibrium point; J0 is the Jacobian matrix at the equilibrium point, each of the above matrices is a constant matrix.

[0078] The discrete state equation and the output equation at time t are established according to formula (3) as

[0079]

[0080] Wherein

[0081]

[0082] In the above formula, T s is a sampling period. The output equation is determined according to the measurement value of the sensor, for example, when the state quantity X is completely measurable, at this time C = I.

[0083] The incremental form of formula is

[0084]

[0085] According to the above equation, the prediction model of future N steps is obtained as

[0086] Y p (t+1|t) = S x Δx(t) + S y y(t) + S u ΔU m (t) (7)

[0087] Wherein, Y p (t+1|t) is the output sequence vector predicted at time t for future N times; ΔU m (t) is the input sequence vector predicted at time t for future m times

[0088]

[0089] The parameter matrices S x , S y , S u are respectively

[0090]

[0091] Step two, according to the prediction model, the output quantity in the prediction time domain is estimated by the control quantity in the control time domain, the optimization target is set according to the control target that the output quantity in the prediction time domain approaches the expected trajectory, the constraint condition is set, and the optimization problem of the control quantity is determined.

[0092] At each time t, the control quantity u(t+k|t), k=0,2,…,m-1 (m is the control time domain) of the future m time points is input into the prediction model as the input quantity, and the output quantity y(t+k|t), k=1,2,…,N (N is the prediction time domain) of the future N time points of the multi-dimensional damping device can be estimated, so as to realize the effect of early prediction.

[0093] The control target of the multi-dimensional damping device is to optimize the control quantity u(t+k|t), k=0,2,…,m-1 of the future m time points, so that the output quantity sequence y(t+k|t), k=1,2,…,N is as close as possible to the expected trajectory r(t), and the control quantity is smooth.

[0094] According to the above control target, the optimization target value is set as

[0095] J(x(t),ΔU m (t))=||Γ y [Y p (t+1|t)-R(t+1)]|| 2 +||Γ u ΔU m (t)|| 2 (10)

[0096] Wherein, R(t+1) is an expected trajectory point vector; Γ y is an output weight matrix; Γ u is an input weight matrix

[0097]

[0098] According to the prediction model of formula, the equivalent target value of the optimization target of formula is obtained

[0099]

[0100] Wherein

[0101]

[0102] The constraints such as stroke constraints, task space limits, actuator power / output limits and the like of the multi-dimensional damping device can be converted into input quantity and output quantity constraints, that is,

[0103]

[0104] The above constraints can be converted to obtain a unified constraint form LΔU m (t)≤b.

[0105] Therefore, the optimization problem can be summarized as

[0106]

[0107] Step three, determine the desired trajectory according to the pose information of the carrier equipment at time t.

[0108] In the multi-dimensional active vibration reduction device, the real-time pose information of the carrier equipment (vehicle or ship) is measured by the inertial navigation unit, and the position and velocity of the vibration reduction device task space corresponding dimension are taken as the input pose r In , and the desired trajectory of the robot task space is set as r(t) = -r In , that is, in the ideal case, the multi-dimensional active vibration reduction device can completely compensate for the vibration disturbance of the vehicle or ship.

[0109] In the actual measurement process, the inertial navigation unit has measurement noise and bias, and a multi-sensor perception fusion algorithm can be designed according to the characteristics of each sensor (accelerometer, gyroscope, magnetometer, GPS, etc.) in the inertial navigation unit to obtain high-precision input pose information.

[0110] Because the model predictive control needs to obtain the information of the desired trajectory r(t) after N time points in advance, the method of obtaining information in advance includes but is not limited to the following methods:

[0111] (1) Install another inertial navigation unit at the front of the vehicle or ship;

[0112] (2) Install a vehicle height sensor on the suspension of the vehicle, and estimate in advance combined with the vehicle dynamics model;

[0113] (3) Install a visual sensor (visual camera, laser radar, millimeter wave radar, etc.) in front of the vehicle to perceive the road surface fluctuation information in advance, and make a prediction combined with the vehicle dynamics model.

[0114] Step four, solve the optimization problem to determine the optimal control sequence, and take the first value in the sequence as the actual control quantity at the current time to control the multi-dimensional vibration reduction device.

[0115] Because the optimization problem of formula is a standard convex optimization problem, it can be solved by active set methods, interior-point methods, gradient projection methods, alternating direction method of multipliers, etc., and the global optimal solution can be determined.

[0116] After solving ΔU m(t), that is, the optimal control quantity sequence u(t+k|t), k=0, 2, …, m-1 at time t, and taking the first value u(t|t) in the sequence as the actual control quantity at the current time to control the multi-dimensional damping device.

[0117] The above steps are repeated at a subsequent time.

[0118] The multi-dimensional damping control method based on model predictive control can realize early detection, prediction of the future, early control, reduction of the influence of system delay, optimal control in a complex constraint environment, and multi-dimensional high-precision damping effect. The control effect can be adjusted by setting the prediction time domain N, the control time domain m, and the target weight parameter.

[0119] In one embodiment, the application provides a multi-dimensional damping control system based on model predictive control, which is applied to a multi-dimensional damping device, and the system comprises:

[0120] A model establishing module is configured to perform kinematics and dynamics analysis on the multi-dimensional damping device, establish a linear dynamics equation near an equilibrium point, and establish a prediction model according to the linear dynamics equation.

[0121] An optimization module is configured to estimate an output quantity in a prediction time domain by a control quantity in a control time domain according to the prediction model, set an optimization target according to a control target that the output quantity in the prediction time domain approaches an expected trajectory, set a constraint condition, and determine an optimization problem of the control quantity.

[0122] A measurement module is configured to determine the expected trajectory according to pose information of the carrying device at time t.

[0123] A control module is configured to solve the optimization problem, determine an optimal control sequence, take the first value in the sequence as the actual control quantity at the current time, and control the multi-dimensional damping device.

[0124] In one embodiment, the application provides a multi-dimensional damping device based on model prediction, as shown in the specification Figure 3 The application provides a multi-dimensional damping device based on model prediction, which comprises a base, an inertial navigation unit, a robot mechanism, and an upper platform.

[0125] The base is fixed on a carrying device, and the carrying device is a vehicle or a ship.

[0126] The inertial navigation unit is installed on the base and is configured to measure pose information of the carrying device. The inertial navigation unit comprises an accelerometer, a gyroscope, a magnetometer, a GPS, etc., and can perceive 6-dimensional pose information of the vehicle or the ship, including displacement, speed and acceleration information of 3-direction movement and 3-direction rotation.

[0127] The upper platform is a platform for carrying people or objects at the end, and is connected to the base through the robot mechanism.

[0128] The robot mechanism has the ability of multi-dimensional active vibration reduction, which can make the upper platform realize high-precision stability and greatly reduce or even avoid the influence of vibration on the personnel or equipment.

[0129] The robot mechanism used can be a parallel robot, a serial robot or a serial-parallel hybrid robot, which internally contains a control unit, a multi-axis servo execution unit (servo electric cylinder, servo hydraulic cylinder, speed reducer linkage, etc.), an encoder and other components, and can realize high-precision, multi-dimensional position and force control. The control unit is electrically connected with the inertial navigation unit, and the control unit includes a memory and a processor, the memory is used to store a control program, and the processor is used to load and execute the control program to realize the multi-dimensional vibration reduction control method based on model predictive control as described above; the execution unit is electrically connected with the control unit and is used to execute operations according to the control instructions of the control unit; the encoder is electrically connected with the control unit and the execution unit, and is used to detect the actual execution of the execution unit and feed back to the control unit. Some of the robot configurations that can be selected are shown in the attached Figure 4 Here, only examples are given, and the control method proposed in the present application is not limited to these several robot configurations.

[0130] In one embodiment, the robot configuration is shown in the attached Figure 5 and the attached Figure 6 A certain type of multi-dimensional active vibration reduction device has four degrees of freedom, i.e. up and down (moving along the y-axis), left and right (moving along the z-axis), pitch (rotating around the z-axis) and roll (rotating around the y-axis). The device is a parallel robot, and the upper platform and the base are connected by four branch chains, of which the first and second branch chains are RPS branch chains, and the third and fourth branch chains are UPS branch chains. The driving pair of each branch chain is the middle sliding pair (P pair), which is driven by a motor electric cylinder. The lower hinge points A1 and A2 are rotary pairs (R pairs); the lower hinge points A3 and A4 are universal joints (U pairs); and the upper hinge points B1, B2, B3 and B4 are spherical hinges (S pairs). At the same time, a spring damper is connected in parallel on each branch chain, which can play a role in load bearing and high-frequency vibration reduction. Therefore, from the classification of vibration reduction principles, the device belongs to a hybrid active and passive vibration reduction device. The size information of the platform and each hinge point is shown in the attached Figure 7 and Table 2, and the body coordinate system O A x A y A z A The body coordinate system O B x B y B z B .

[0131] Table 2 Parameters of the multi-dimensional active vibration reduction device of this type

[0132]

[0133] The upper platform body coordinate system O B x B y B z B The displacement amount (task space coordinates) relative to the base body coordinate system O A x A y A z A

[0134]

[0135] wherein y is the relative displacement amount up and down; z is the relative displacement amount left and right; a is the relative roll angle; and g is the relative pitch angle. Under the definition of the rated load, the balance point of the device is

[0136]

[0137] At this balance point, the balance height of the device is 400 mm, and there is no relative displacement in the left-right, roll and pitch directions. At this time, the length of each branch chain is the same, which is the balance length, that is,

[0138]

[0139] During operation, the multi-dimensional active vibration reduction device moves around the balance point, and the movement range in each direction of the task space is

[0140] -50mm≤y-y0≤50mm,-20mm≤z≤20mm,-5°≤a≤5°,-8°≤g≤8° (19)

[0141] Therefore, the nonlinear dynamic equation can be linearized at this balance point, and the linearized dynamic equation is

[0142]

[0143] wherein u1, u2, u3 and u4 are the control forces applied by the electric cylinders on each branch chain; J0 is the Jacobian matrix at the balance point

[0144]

[0145] M0 is the mass matrix at the balance point

[0146]

[0147] In the above formula, m = 100 kg is the equivalent mass of the system at the balance point; J x0 = 2 kg·m 2 ​is the equivalent moment of inertia of the system around the x-axis at the equilibrium point; J z0 =5kg·m 2 is the equivalent moment of inertia of the system around the z-axis at the equilibrium point.

[0148] When the inertial navigation unit of the multi-dimensional active vibration reduction device base measures the posture information of the vehicle or ship, the input posture quantity is

[0149] r In =[y r z r α r γ r ] T (twenty three)

[0150] The purpose of the multi-dimensional active vibration reduction device is to reduce vibration in all directions as much as possible. Therefore, the device should actively compensate for the posture disturbance of the vehicle or ship. That is, the control goal is to make the system state quantity X track the dynamic trajectory r = -r In The optimization goal is to make the system tracking error as small as possible, while the control amount should be relatively smooth, and it needs to meet the device's motion range limitation (formula) and the electric cylinder's output limitation.

[0151] -500N≤u1,u2,u3,u4≤500N (24)

[0152] According to the multi-dimensional vibration reduction control method proposed in the present invention, model predictive control is used to solve the above-mentioned optimal control problem with multiple constraints.

[0153] Set the sampling period to 0.01s, the prediction time domain to 10, the control time domain to 2, the output weight to 20, the input weight to 0, the input increment weight to 0.1, set the constraints according to the formula and the formula, and normalize all input and output variables.

[0154] Assume that the vehicle posture information measured by the inertial navigation unit is as follows

[0155]

[0156] The final result is as shown in the attached manual. Figure 7 and attached Figure 8 As shown. Figure 7 It can be seen that the output of the electric cylinder of each branch chain is within the range of 500N, which meets the requirements of the input constraint. Figure 8 It can be seen that the output of the multi-dimensional active vibration reduction device can track the desired dynamic trajectory very well, so it can achieve a good vibration reduction effect in the up and down, left and right, pitch and roll directions. Figure 8It can be seen that all output quantities are within the motion range of the device, satisfying the output constraint condition, in particular, the amplitude of the left and right input excitation reaches 0.03m in the second figure, and the response amplitude of the output is 0.02m, which does not exceed the motion range limit (the vibration amount exceeding the motion range cannot be damped). It can be seen that the multi-dimensional damping control method based on model predictive control proposed in the application can well deal with the efficient damping problem under the actual constraint condition.

[0157] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.

[0158] Those skilled in the art can appreciate that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized by hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0159] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e. they can be located in one place or distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiment according to actual needs.

[0160] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0161] In the present application, the term "comprise", "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the elements defined by the statement "comprise" do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0162] It should be noted that the above embodiments can be freely combined as needed. The above are only preferred embodiments of the present application, and it should be pointed out that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should also be considered as the protection scope of the present application.

Claims

1. A multi-dimensional vibration reduction control method based on model predictive control, characterized in that: Applied to a multi-dimensional vibration reduction device, the method comprises the following steps: S100: Perform kinematic and dynamic analysis on the multi-dimensional vibration reduction device, establish a linear dynamic equation near the equilibrium point, and establish a prediction model based on the linear dynamic equation; S200: Estimate the output of the prediction time domain by using the control quantity of the control time domain according to the prediction model, set the optimization target according to the control target that the output of the prediction time domain approaches the desired trajectory, set the constraint conditions, and determine the optimization problem of the control quantity; S300: Determine the desired trajectory based on the position information of the carrier at time t; S400: Solve the optimization problem, determine the optimal control variable sequence, and use the first value in the sequence as the actual control variable at the current moment to control the multi-dimensional vibration reduction device; S500: t=t+1, return to S300; The prediction model established according to the linear kinetic equation in S100 is: Y p (t+1|t)=S x Δx(t)+S y y(t)+S u ΔU m (t) Among them, Y p (t+1|t) is the output sequence vector predicted at time t for the next N moments; ΔU m (t) is the input sequence vector predicted at time t for m moments in the future; S x 、S y 、S u is the parameter matrix; Δx(t) is the increment of the state variable at time t; y(t) is the system output at time t; The optimization problem in S200 is determined as: s.t.LΔU m (t)≤b Among them, S u is the parameter matrix; is the parameter matrix transpose; R(t+1) is the expected trajectory point vector; Γ y is the output weight matrix; Γ u is the input weight matrix.

2. The multi-dimensional vibration reduction control method based on model predictive control according to claim 1, characterized in that: The linear dynamic equation near the equilibrium point in S100 is: in, is the task space coordinate; is the task space velocity; M0 is the task space mass matrix at the equilibrium point; C0 is the task space damping matrix at the equilibrium point; K0 is the task space stiffness matrix at the equilibrium point; J0 is the Jacobian matrix at the equilibrium point; u is the control force.

3. The multi-dimensional vibration reduction control method based on model predictive control according to claim 1, characterized in that: The step S200 of estimating the output of the prediction time domain by using the control amount of the control time domain according to the prediction model, and setting the optimization target according to the control target that the output of the prediction time domain approaches the desired trajectory includes: S201: Determine the optimization target J(x(t),ΔU m (t))=||Γ y [Y p (t+1|t)-R(t+1)]|| 2 +||C u D.U. m (t)|| 2 Where, J(x(t),ΔU m (t)) is the optimization objective function at time t; Y p (t+1|t) is the output sequence vector predicted at time t for the next N moments; ΔU m (t) is the input sequence vector predicted at time t for m moments in the future; R(t+1) is the expected trajectory point vector; Γ y is the output weight matrix; Γ u is the input weight matrix; S202: Obtaining an equivalent target value of the optimization target according to the prediction model: in, is the equivalent optimization objective function at time t; S u is the parameter matrix; is the transpose of the parameter matrix.

4. The multi-dimensional vibration reduction control method based on model predictive control according to claim 3, characterized in that: The constraints in S200 include stroke limitation, motion space limitation, and actuator power / output limitation.

5. The multi-dimensional vibration reduction control method based on model predictive control according to claim 1, characterized in that: Determining the expected trajectory according to the position information of the carrier at time t in S300 includes: The inertial navigation unit is used to measure the posture information of the carrier at time t, and the position and velocity of the corresponding dimension of the vibration reduction device task space are used as the input posture quantity r In , and set the expected trajectory to r(t) = -r In .

6. The multi-dimensional vibration reduction control method based on model predictive control according to claim 5, characterized in that: After the inertial navigation unit is used to measure the position and posture information of the carrier at time t, the method further includes: filtering the obtained position and posture information through a multi-sensor perception fusion algorithm.

7. A multi-dimensional vibration reduction control system based on model predictive control, characterized in that: Applied to a multi-dimensional vibration reduction device, the system comprises: Model building module, used to perform kinematic and dynamic analysis on the multi-dimensional vibration reduction device, establish linear dynamic equations near the equilibrium point, and establish a prediction model based on the linear dynamic equations; The optimization module is used to estimate the output quantity in the prediction time domain through the control quantity in the control time domain according to the prediction model, set the optimization target according to the control target that the output quantity in the prediction time domain approaches the expected trajectory, set the constraint conditions, and determine the optimization problem of the control quantity; The measurement module is used to determine the desired trajectory based on the position information of the vehicle at time t; The control module is used to solve the optimization problem, determine the optimal control sequence, and use the first value in the sequence as the actual control variable at the current moment to control the multi-dimensional vibration reduction device; Wherein, the prediction model established according to the linear kinetic equation is: Y p (t+1|t)=S x Δx(t)+S y y(t)+S u ΔU m (t) Among them, Y p (t+1|t) is the output sequence vector predicted at time t for the next N moments; ΔU m (t) is the input sequence vector predicted at time t for m moments in the future; S x 、S y 、S u is the parameter matrix; Δx(t) is the increment of the state variable at time t; y(t) is the system output at time t; The optimization problem is determined as: s.t.LΔU m (t)≤b Among them, S u is the parameter matrix; is the parameter matrix transpose; R(t+1) is the expected trajectory point vector; Γ y is the output weight matrix; Γ u is the input weight matrix.

8. A multi-dimensional vibration reduction device based on model prediction, characterized in that: include: A base, fixed to the carrier; An inertial navigation unit, mounted on the base, for measuring the position and posture information of the carrier; The robot mechanism includes a control unit, an execution unit, and an encoder, wherein: The control unit is electrically connected to the inertial navigation unit, and the control unit includes a memory and a processor, the memory is used to store a control program, and the processor is used to load and execute the control program to implement the multi-dimensional vibration reduction control method based on model predictive control according to any one of claims 1 to 6; The execution unit is electrically connected to the control unit and is used to execute operations according to control instructions of the control unit; The encoder is electrically connected to the control unit and the execution unit, and is used to detect the actual execution status of the execution unit and feed back to the control unit; The upper platform is connected to the base through the robot mechanism and is used for carrying people or objects.

Images