Volume quantitative control method of retinal intravenous injection system

Abstract

The invention discloses a volume quantitative control method of a retina intravenous injection system, which belongs to the technical field of ophthalmology robots, and comprises the following steps: firstly, designing a smooth injection volume curve as an expected volume of the quantitative control method; aiming at disturbance characteristics of different stages of the injection system, an evolution observer is designed to estimate unknown disturbance and compensate the unknown disturbance into a controller; an equivalent control law designed by adopting an integral sliding mode surface and an evolution observer is combined with a switching control law designed by adopting a self-adaptive gain, and an obtained total control law can realize reliable tracking of an expected volume. The method can ensure high-precision closed-loop control of the drug volume in the retinal intravenous injection operation, and has robust tracking performance.

Description

Technical Field

[0001] This invention belongs to the field of ophthalmic robotics technology, specifically relating to a method for quantitative volume control of a retinal intravenous injection system. Background Technology

[0002] Retinal intravenous injection surgery allows for precise drug delivery via blood vessels, particularly by injecting thrombolytic drugs into occluded retinal veins to restore local blood circulation and reduce the risk of blindness. However, current retinal intravenous injection surgeries often use a viscous fluid control unit on a vitrectomy machine to drive the drug injection. This process uses a fixed pressure, and the initial burst of high flow rate can damage local blood vessels.

[0003] Considering the side effects of the drug, an excessive dose can increase the risk of bleeding in local tissues; while an insufficient dose can weaken the expected therapeutic effect. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing a method for quantitative volume control of a retinal intravenous injection system. This method ensures precise tracking of the desired volume during retinal intravenous injection, guarantees the drug delivery effect, and reduces surgical risks.

[0005] This invention achieves this objective through the following technical solution: A method for volume-based quantitative control of a retinal intravenous injection system, comprising the following: A third-order dynamic model of injection volume was constructed for the retinal vein injection system to determine the input and output variables of the injection system. Design a smooth injection volume curve as the desired output variable for the volume quantification control method; An evolutionary observer was designed to address unknown perturbations in the injection system. This observer comprises two parts: an initial observer and a steady-state observer, which switch between them at preset intervals. The steady-state observer specifically includes a mechanism for estimating the perturbation derivative. This observer ensures both safety during the initial injection phase and accuracy during the steady-state phase.

[0006] A control law is designed using a fixed-time integral sliding surface, including an equivalent control law based on the sliding surface and an evolutionary observer, and a switching control law designed with adaptive gain. The total control law is obtained by combining the equivalent control law and the switching control law to control the input variables in the volume third-order dynamic model of the retinal intravenous injection system.

[0007] The specific details of constructing the third-order kinetic model of the injection volume are as follows: The injection system uses an electric motor-driven piston. P 1. To push the silicone oil, which in turn pushes the syringe piston. P2. Finally, the medication in the syringe is injected into the retinal vein cavity; piston P The first-order kinetics of the process driving the flow of silicone oil are described as follows: ; in and These represent the silicone oil flow rate and its first derivative, respectively. Indicates piston P The speed of 1 This indicates an unknown disturbance in the process. , and These are positive parameters; syringe piston P speed of 2 With silicone oil flow The relationship is: ; in For syringe piston P The area of ​​2, This indicates the unknown dynamics of the process; By syringe piston P The drug flow dynamics driven by 2 can be described as follows: ; in and Let their respective values ​​represent the drug flow rate and its first derivative. Indicates piston P The speed of 2 This indicates an unknown disturbance in the process. , and These are positive parameters; Based on drug injection volume The continuity of is described by its first-order dynamics as follows: ; in The first derivative of the drug volume; Based on the above four formulas, the volumetric third-order dynamic model of the retinal vein injection system is obtained: ; in , , , The third derivative of the drug volume. The second derivative of the drug volume. The first derivative of the drug volume. For aggregated disturbances, ,in for The first derivative; Selecting state variables , As the control input, the third-order dynamic model is rewritten as follows: ; By designing a closed-loop controller to control the drive motor to push the piston. P The speed of 1 is used to control the drug volume.

[0008] Furthermore, the injection volume curve, which serves as the desired output variable, is as follows: set up A fixed-flow injection strategy will be adopted initially over a specific time period. The time interval uses a quadratic curve to ensure a smooth transition of the trajectory until the final set volume is reached. Stop time for constant flow state This refers to the injection stop time; Based on the principle of continuity between flow rate and volume, the following conditions are obtained: ; in yes Expected drug volume at time, yes Expected drug volume at time, Is the volume curve in The first derivative, yes The first derivative of the expected drug volume at any given time is obtained by solving the above equation: ; The desired injection volume is: .

[0009] Furthermore, an evolutionary observer is designed to estimate the lumped uncertainty of the system, as follows: In the pre-setting stage The following initial observers are used to estimate the disturbance over the time period: ; in for Estimates of the output system uncertainty of the initial observer over the time period. , , , It is a positive parameter. and For parameters greater than 1, and For parameters that are greater than 0 and less than 1, As an auxiliary variable; In the pre-setting stage A steady-state observer is used to estimate the disturbance over a given time period. The unknown disturbance dynamics are assumed to be... ,in Given an unknown nonlinear mapping function, based on the Koopman operator and the extended dynamic mode decomposition method, we approximate the unknown perturbation dynamics using a finite-dimensional linear system. First, we construct a lifting function vector based on the unknown perturbation autoregressive vector and the state variables: ; in Let the order be the autoregressive order. Let be the state variable vector at time t; Unknown perturbation dynamics The linear model in the lifting function space is described in the following discrete-time form: ; in To approximate the Koopman matrix, the vectors in this matrix... Solving using linear Bayesian regression: ; in , , for The One element, For data volume; It is a positive parameter. It is the identity matrix; Transform the discrete system model with unknown disturbances into a continuous system model to match the design of the steady-state observer: ; in , Sampling time, It is the identity matrix. ; In preset The steady-state observer for the time period is designed as follows: ; in for Estimates of the output system uncertainty of the steady-state observer over the time period. For matrix The first row vector, For vectors The posterior standard deviation; For adaptive gain, its update law is: ; in It is a positive number.

[0010] Furthermore, the integral sliding surface used is represented as follows: ; in , , , , To track errors, State vector The i One element, For the desired state variable, , , and For the preset threshold, For symbolic functions, It is the integral variable.

[0011] Furthermore, the designed equivalent control law for: ; in Let be the third derivative of the desired volume.

[0012] Furthermore, switch control laws Designed as follows: ; in For adaptive gain switching, It is a positive parameter. It is a constant. , .

[0013] Furthermore, adaptive gain switching The adaptive update law is: ; in It is a positive number.

[0014] Furthermore, data within a fixed time window is used. Solve for the approximate Koopman matrix.

[0015] Compared with the prior art, the beneficial effects of this invention are as follows: The volume control method for a retinal intravenous injection system provided by this invention can control the drug volume in real time according to the set desired volume through flow feedback. This not only improves the instability of the initial injection process under pressure drive, but also improves the tracking accuracy of drug volume during injection. Simultaneously, it has good adaptive capability, effectively improving the robustness of the control process. Attached Figure Description

[0016] Figure 1 The retinal vein injection system model provided by the present invention; Figure 2 This is a schematic diagram of the desired injection volume provided by the present invention; Figure 3 Flowchart of the evolutionary observer implementation provided by this invention; Figure 4 The flowchart of the injection system volume quantitative control method provided by the present invention is shown below. Figure 5 This is a diagram showing the effect of tracking injection volume obtained from experiments provided by the present invention; Figure 6 The diagram shows the effect of tracking injection flow rate obtained from the experiment provided by this invention. Detailed Implementation

[0017] Exemplary embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the invention to those skilled in the art. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0018] The specific modeling method for the retinal vein injection system is as follows: like Figure 1 As shown, the injection system uses a motor-driven piston. P 1. To push the silicone oil, which in turn pushes the syringe piston. P 2. Finally, the drug is injected into the retinal vein cavity; piston P The first-order kinetics of the process driving the flow of silicone oil are described as follows: ; in and These represent the silicone oil flow rate and its first derivative, respectively. Indicates piston P The speed of 1 This indicates an unknown disturbance in the process. , and These are positive parameters; syringe piston P speed of 2 With silicone oil flow The relationship is: ; in For syringe piston P The area of ​​2, This indicates the unknown dynamics of the process; By syringe piston P The drug flow dynamics driven by 2 can be described as follows: ; in and Let their respective values ​​represent the drug flow rate and its first derivative. Indicates piston P The speed of 2 This indicates an unknown disturbance in the process. , and These are positive parameters; Based on drug injection volume The continuity of is described by its first-order dynamics as follows: ; in It is the first derivative of the drug volume.

[0019] Based on the above four formulas, the volumetric third-order dynamic model of the retinal vein injection system is obtained: ; in , , , The third derivative of the drug volume. The second derivative of the drug volume. The first derivative of the drug volume. For aggregated disturbances, ,in For syringe piston P 2. Unknown disturbances related to drive The first derivative; Selecting state variables , As the control input, the third-order dynamic model is rewritten as follows: ; As can be seen from the above equation, by designing a closed-loop controller to control the drive motor to push the piston... P The speed of 1 is used to control the drug volume.

[0020] The specific method for planning the expected injection volume is as follows: like Figure 2 As shown, let A fixed-flow injection strategy will be adopted initially over a specific time period. The time interval uses a quadratic curve to ensure a smooth transition of the trajectory until the final set volume is reached. Among these, Stop time for constant flow state This refers to the injection stop time. By employing an early deceleration strategy, over-injection of medication is prevented while ensuring smooth flow output. Based on the principle of continuity between flow rate and volume, the following conditions can be obtained: ; in yes Expected drug volume at time, yes Expected drug volume at time, Is the volume curve in The first derivative, that is Constant drug flow rate over a period of time yes The first derivative of the expected drug volume at any given time is obtained by solving the above equation: ; The final desired injection volume is: ; Therefore, at a known time Final required injection volume and initial traffic Under the premise of a smooth tracking curve It can be uniquely determined.

[0021] The evolutionary observer design process used in this invention is as follows: Design an evolutionary observer to estimate the lumped uncertainty of the system. Specifically, in the pre-set phase... The following initial observer is used to estimate the disturbance during the time period to ensure safety in the initial stage: ; in, The transient time is a pre-set value, generally determined based on the transient time results of the drug flow rate. for Estimates of the output system uncertainty of the initial observer over the time period. , , , It is a positive parameter. and For parameters greater than 1, and For parameters that are greater than 0 and less than 1, As an auxiliary variable, it avoids the noise amplification problem caused by numerical differentiation.

[0022] In the pre-setting stage A steady-state observer is used to estimate the disturbance over a given time period to ensure accuracy during the steady-state phase. The unknown disturbance dynamics are assumed to be... ,in Let be an unknown nonlinear mapping function. Based on the Koopman operator and the extended dynamic mode decomposition method, the dynamics of unknown perturbations can be approximated using a finite-dimensional linear system. First, a lifting function vector based on the unknown perturbation autoregressive vector and the state variables is constructed: ; in Let the order be the autoregressive order. Let be the state variable vector at time t. Therefore, the unknown disturbance dynamics... The linear model in the lifting function space can be described in the following discrete-time form: ; in , is an approximate Koopman matrix, in which vectors Solving using linear Bayesian regression: ; in , , for The One element, For data volume. It is a positive parameter. It is an identity matrix. This invention uses data within a fixed time window. Solve for the approximate Koopman matrix.

[0023] Transform the discrete system model with unknown disturbances into a continuous system model to match the design of the steady-state observer: ; in , Sampling time, It is the identity matrix. .

[0024] Therefore, in the preset The steady-state observer for the time period is designed as follows: ; in for Estimates of the output system uncertainty of the steady-state observer over the time period. For matrix The first row vector, For vectors The posterior standard deviation, For adaptive gain, its update law is: ; in It is a positive number.

[0025] Figure 3 The flowchart of the evolutionary observer design is shown. The evolutionary observer enables the injection system to specifically overcome the transient piston response in the initial stage and the unknown disturbances encountered in the steady-state stage, thereby ensuring the volume tracking accuracy of the injection system.

[0026] The control algorithm is as follows: The integral sliding surface used in this invention is represented as follows: ; in , , , , To track errors, State vector The i One element, For the desired state variable, and It is a constant. , , and For the preset threshold, For symbolic functions, It is the integral variable.

[0027] Based on the third-order volume dynamics model of the injection system, the integral sliding surface, and the evolutionary observer ( 0~t 0 As the initial observer, t 0 ~+∞ The equivalent control law designed for a steady-state observer for: ; in For the desired volume The third derivative of .

[0028] The switching control law designed in this invention Designed as follows: ; in For adaptive gain switching, It is a positive parameter. It is a constant, where , .

[0029] The adaptive update law is: ; in It is a positive number.

[0030] Figure 4 A flowchart illustrating the volumetric injection control method for the injection system is presented. This volumetric injection control method effectively avoids initial process instability and low steady-state tracking accuracy caused by unknown external disturbances.

[0031] A tracking and control experiment for the desired volume was conducted based on the established retinal intravenous injection system. For example... Figure 5 As shown, for the desired volume curve, upon reaching... Inject at a constant flow rate of 96 μL / min 50 s before the time limit. At 50 s, the desired injection volume reaches 80 μL, at which point the flow rate begins to decrease, and the final desired injection volume... =100μL. It can be seen that the method of this invention can track to the vicinity of the desired volume more quickly, with smaller errors and higher steady-state accuracy. Figure 6 The control method of the present invention demonstrates that it can effectively reduce the piston transient response in the initial stage, with less overshoot and better safety.

[0032] The present invention has been described in detail above through embodiments, but the content described is only an exemplary embodiment of the present invention and should not be considered as limiting the scope of the present invention. The scope of protection of the present invention is defined by the claims. Any technical solutions designed by those skilled in the art using the technical solutions described in the present invention, or similar technical solutions designed by those skilled in the art under the inspiration of the technical solutions of the present invention, within the substance and scope of protection of the present invention, to achieve the above-mentioned technical effects, or equivalent changes and improvements made to the scope of the application, should still fall within the patent protection scope of the present invention. It should be noted that, for clarity, descriptions of some components and processes that are not directly and obviously related to the scope of protection of the present invention but are known to those skilled in the art have been omitted in the description of the present invention.

Claims

1. A method for quantitative volume control of a retinal intravenous injection system, characterized in that, Includes the following: A third-order dynamic model of injection volume was constructed for the retinal vein injection system to determine the input and output variables of the injection system. Design a smooth injection volume curve as the desired output variable for the volume quantification control method; An evolutionary observer is designed for unknown perturbations in the injection system. The evolutionary observer consists of two parts: an initial observer and a steady-state observer, which are switched at preset times. The steady-state observer specifically includes a mechanism for estimating the perturbation derivative. A control law is designed using a fixed-time integral sliding surface, including an equivalent control law based on the sliding surface and an evolutionary observer, and a switching control law designed with adaptive gain. The total control law is obtained by combining the equivalent control law and the switching control law to control the input variables in the volume third-order dynamic model of the retinal intravenous injection system.

2. The method according to claim 1, characterized in that, The specific details of constructing the third-order kinetic model of the injection volume are as follows: The injection system uses an electric motor-driven piston. P 1. To push the silicone oil, which in turn pushes the syringe piston. P 2. Finally, the medication in the syringe is injected into the retinal vein cavity; piston P The first-order kinetics of the process driving the flow of silicone oil are described as follows: ； in and These represent the silicone oil flow rate and its first derivative, respectively. Indicates piston P The speed of 1 This indicates an unknown perturbation in the process. , and These are positive parameters; syringe piston P speed of 2 With silicone oil flow The relationship is: ； in Let be the area of ​​the syringe piston. This indicates the unknown dynamics of the process; By syringe piston P The drug flow dynamics driven by 2 can be described as follows: ； in and Let their respective values ​​represent the drug flow rate and its first derivative. Indicates piston P The speed of 2 This indicates an unknown perturbation in the process. , and These are positive parameters; Based on drug injection volume The continuity of is described by its first-order dynamics as follows: ； in The first derivative of the drug volume; Based on the above four formulas, the volumetric third-order dynamic model of the retinal vein injection system is obtained: ； in , , , The third derivative of the drug volume. The second derivative of the drug volume. The first derivative of the drug volume. For aggregated disturbances, ,in for The first derivative; Selecting state variables , As the control input, the third-order dynamic model is rewritten as follows: ； By designing a closed-loop controller to control the drive motor to push the piston. P The speed of 1 is used to control the drug volume.

3. The method according to claim 2, characterized in that, The injection volume curve, which is the expected output variable, is as follows: set up A fixed-flow injection strategy will be adopted initially over a specific time period. The time interval uses a quadratic curve to ensure a smooth transition of the trajectory until the final set volume is reached. Stop time for constant flow state This refers to the injection stop time; Based on the principle of continuity between flow rate and volume, the following conditions are obtained: ； in yes Expected drug volume at time, yes Expected drug volume at time, Is the volume curve in The first derivative, yes The first derivative of the expected drug volume at any given time is obtained by solving the above equation: ； The desired injection volume is: 。 4. The method according to claim 3, characterized in that, Design an evolutionary observer to estimate the lumped uncertainty of the system, as follows: In the pre-setting stage The following initial observers are used to estimate the disturbance over the time period: ； in for Estimates of the output system uncertainty of the initial observer over the time period. , , , It is a positive parameter. and For parameters greater than 1, and For parameters that are greater than 0 and less than 1, As an auxiliary variable; In the pre-setting stage A steady-state observer is used to estimate the disturbance over a given time period. The unknown disturbance dynamics are assumed to be... ,in Given an unknown nonlinear mapping function, based on the Koopman operator and the extended dynamic mode decomposition method, we approximate the unknown perturbation dynamics using a finite-dimensional linear system. First, we construct a lifting function vector based on the unknown perturbation autoregressive vector and the state variables: ； in Let the order be the autoregressive order. Let be the state variable vector at time t; Unknown disturbance dynamics The linear model in the lifting function space is described in the following discrete-time form: ； in To approximate the Koopman matrix, the vectors in this matrix... Solving using linear Bayesian regression: ； in , , for The One element, For data volume, It is a positive parameter. It is the identity matrix; Transform the discrete system model with unknown disturbances into a continuous system model to match the design of the steady-state observer: ； in , Sampling time, It is the identity matrix. ; In preset The steady-state observer for the time period is designed as follows: ； in for Estimates of the output system uncertainty of the steady-state observer over the time period. For matrix The first row vector, For vectors The posterior standard deviation; For adaptive gain, its update law is: ； in It is a positive number.

5. The method according to claim 4, characterized in that, The integral sliding surface used is represented as follows: ； in , , , , To track errors, State vector The One element, For the desired state variable, , , and For the preset threshold, For symbolic functions, It is the integral variable.

6. The method according to claim 5, characterized in that, The designed equivalent control law for: ； in Let be the third derivative of the desired volume.

7. The method according to claim 6, characterized in that, Switching control laws Designed as follows: ； in For adaptive gain switching, It is a positive parameter. It is a constant. , .

8. The method according to claim 7, characterized in that, Adaptive switching gain The adaptive update law is: ； in It is a positive number.

9. The method according to claim 4, characterized in that, Data within a fixed time window Solve for the approximate Koopman matrix.

Images