CN114740728A - Isolated heart perfusion blood pressure self-adaptive control method based on semi-parametric model - Google Patents

Isolated heart perfusion blood pressure self-adaptive control method based on semi-parametric model Download PDF

Info

Publication number
CN114740728A
CN114740728A CN202210439215.8A CN202210439215A CN114740728A CN 114740728 A CN114740728 A CN 114740728A CN 202210439215 A CN202210439215 A CN 202210439215A CN 114740728 A CN114740728 A CN 114740728A
Authority
CN
China
Prior art keywords
model
isolated heart
heart
semi
blood
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202210439215.8A
Other languages
Chinese (zh)
Other versions
CN114740728B (en
Inventor
姚蔚然
辛立明
吴承伟
吴立刚
杜德嵩
楚士杰
李通
宋海旭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Harbin Institute of Technology
University of Shanghai for Science and Technology
Original Assignee
Harbin Institute of Technology
University of Shanghai for Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Harbin Institute of Technology, University of Shanghai for Science and Technology filed Critical Harbin Institute of Technology
Priority to CN202210439215.8A priority Critical patent/CN114740728B/en
Publication of CN114740728A publication Critical patent/CN114740728A/en
Application granted granted Critical
Publication of CN114740728B publication Critical patent/CN114740728B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance

Abstract

An in vitro heart perfusion blood pressure self-adaptive control method based on a semi-parametric model belongs to the technical field of in vitro organ protection intelligent systems. The problems that the existing isolated heart perfusion technology has heart damage, the blood pressure regulation is difficult to rapidly stabilize, and the heart can be damaged when the current controller adjusts the gain due to the changed heart perfusion condition are solved. The invention models the isolated heart by combining a parametric model and a nonparametric model, adaptively changes the internal parameters of the model according to the change of each parameter of the isolated heart, generates a reference control track on an isolated heart reference model by a virtual controller, updates control gain by using a target average aortic pressure, an actual average aortic pressure, an isolated heart semi-parametric model and the virtual controller, and controls the flow rate of perfusion blood by a control gain controller. The method is mainly used for self-adaptive control of isolated heart perfusion blood pressure.

Description

Isolated heart perfusion blood pressure self-adaptive control method based on semi-parametric model
Technical Field
The invention relates to an in vitro heart perfusion blood pressure self-adaptive control method, and belongs to the technical field of in vitro organ protection intelligent systems.
Background
The heart transplantation is the gold method for solving the problem of the end stage of heart failure, and is vital to the heart transplantation for the in vitro preservation of the heart. The heart used for transplantation is intended to maintain intact anatomy and viability during transplantation from a donor to a recipient. However, the loss of oxygen and nutrient supply to the isolated heart once removed from the donor accelerates death. The traditional heart supplying method adopts a low-temperature cold preservation method to delay the death rate of the heart, but the preservation time is only 4-6 hours, and the death rate after transplantation is very high after the preservation time exceeds the time range. The isolated heart perfusion technique is a new generation of donor heart preservation method, which maintains the normal metabolism of an extracorporeal donor heart by using blood containing oxygen and rich in nutrition at physiological temperature so as to maintain the activity of the donor heart.
In the process of isolated heart perfusion, the control of the pressure of the aorta is crucial to the maintenance of the normal metabolism of the cardiac muscle cells, and the large fluctuation of the blood pressure can influence the metabolism of the heart, so that the arrhythmia causes the irreversible damage of the cardiac cells. The difficulty in controlling the aortic pressure is that the organs of the body are complex and difficult to control. For example, the resistance of the coronary vessels varies with perfusion time and temperature, and especially in the early stages of blood perfusion, a rapid decrease in coronary vessel resistance can lead to a rapid decrease in aortic blood pressure. At the same time, the regulation of the blood pressure of the heart must be rapid and stable due to the vulnerability of the heart. Therefore, the traditional model-free control method is difficult to achieve precise control of the blood pressure. In addition, the dynamic response of the heart beat typically varies significantly with temperature, drug dose, different heart, etc., which presents a significant challenge to the control of the perfusion blood pressure of the heart. To improve performance in this case, the controller needs to selectively adjust the control gain level according to different ex vivo cardiac characteristics, but these trial-and-error based parameter adjustments may result in irreversible cardiac damage. In order to solve the problem, the invention provides an isolated heart perfusion blood pressure self-adaptive control method based on a semi-parameterized model.
Disclosure of Invention
The invention aims to solve the problems that in the existing extracorporeal heart perfusion technology, the heart is damaged due to the change of various parameters of an extracorporeal heart model, the blood pressure regulation is difficult to quickly stabilize, and the heart is damaged when a current controller adjusts gain due to the changed perfusion condition of the heart.
A method of modeling an ex vivo heart, comprising the steps of:
firstly, establishing a parameter model in an isolated heart semi-parameter model, wherein the parameter model is a ternary model, the ternary model describes the relation between the aortic pressure and the perfusion flow by using a capacitor and two resistors, and the resistor R ismRear parallel connection resistor RdAnd a capacitor Cm，RmIs the flow resistance of blood through the coronary arteries, CmIs the volume of flow of the coronary artery, RdIs the flow resistance of blood flowing through the distal coronary artery;
then determining the impedance of the aorta according to kirchhoff's law based on the ternary model;
the parametric model is represented as:
Figure BDA0003614387940000021
Figure BDA0003614387940000022
wherein the content of the first and second substances,
Figure BDA0003614387940000023
a model of a parameter is represented by,
Figure BDA0003614387940000024
representing inputs of a parametric model, qp(t + Δ t) represents the output of the parametric model;
describing nonlinearities and uncertainties contained in the ex vivo heart model using a non-parametric model; the learning data set is represented as:
Figure BDA0003614387940000025
wherein x isi、yiInput data x (t) and output data y (t) in the learning data set; n is the learning dataset size;
x(t)＝[qp(t)，pa(t)，pa(t-Δt)，pa(t-2Δt)…pa(t-DΔt)]∈RD+2
y(t)＝pa(t+Δt)∈R
wherein q ispA flow rate indicative of a flow rate at which the blood flow rate control unit perfuses blood into the isolated heart; p is a radical ofaThe aortic pressure of the isolated heart is represented, and delta t represents the sampling time interval corresponding to the data; d represents the dimension of the input;
the sampled input data follows a Gaussian process distribution, and a non-parametric model f (x) is expressed as follows:
f(x)～GP(μ(x)，k(x，x′))
wherein GP (·) represents a Gaussian distribution; μ (x) represents the mean function of the input data x (t), k (x, x ') represents the covariance function of any two input data x and x';
μ(x)＝E[f(x)]
k(x，x′)＝E[(f(x)-μ(x))(f(x′)-μ(x′))]
wherein E [. cndot. ] represents a mathematical expectation; using parametric models as a priori knowledge, i.e.
Figure BDA0003614387940000026
Figure BDA0003614387940000027
Wherein
Figure BDA0003614387940000028
The semi-parameter model of the isolated heart is f (x) -GP (G (xA), k (x, x')).
Further, based on the ternary model, the aorta impedance determined according to kirchhoff's law is as follows:
pa-pm＝Rm·qa
Cm·dpm/dt＝qm
pm＝Rd·(qa-qm)
wherein p isaShowing the actual measured isolated heart aortic pressure, qaIs the real aortic flow of the isolated heart obtained by actual measurement; p is a radical ofmIs RmPressure at the rear node; q. q ofmIs the blood passing through CmThe flow rate of (c).
Further, the input data x (t) and the output data y (t) in the learning data set are data in an optimal parameter set determined according to the sampling data, and the process of determining the optimal parameter set according to the sampling data comprises the following steps:
firstly, acquiring sampling data, wherein the sampling data is the flow rate q for perfusing blood into an isolated heartpAortic pressure p of isolated hearta；
Then, a sampling data set is constructed by using the sampling data: the flow rate q of the blood perfused into the isolated heart by the blood flow rate control unit at the current momentpThe aortic pressure of the isolated heart at the current momentAnd the aortic pressure p of the isolated heart at the past D momentsaAs input x (t); taking the aortic pressure of the isolated heart perfused by the centrifugal pump at the next sampling moment as output y (t);
based on the sampled data set, the parameter configuration of the parameterized model is recorded as S ═ Rm，Cm，RdAdjusting parameters in S in a perturbation mode to enable the minimum root mean square error RMSE between the actual measured aortic pressure and the predicted aortic pressure to obtain an optimal parameter set S*。
Further, based on the semi-parametric model of the isolated heart, a point x is queried for the model*The predicted results are as follows:
Figure BDA0003614387940000031
Figure BDA0003614387940000032
wherein
y＝[y1，…，yN]T
Figure BDA0003614387940000033
μ(X)＝[μ(x1)，…，μ(xN)]T
K＝k(X，X)
Km，n＝k(xm，xn)
k*＝k(X，x*)
km，*＝k(xm，x*)
k**＝k(x*，x*)
Wherein σnIs x*I is a unit matrix.
The self-adaptive control method of the perfusion blood pressure of the isolated heart based on the semi-parametric model comprises the following steps:
the method comprises the steps that a semi-parameter model of the isolated heart established based on an isolated heart modeling method is tested by utilizing an open-loop step response mode in a Ziegler-Nichols method, and then parameters of a virtual PID controller are obtained by utilizing an overshoot-free rule of the Ziegler-Nichols method;
maintaining vital and optimally functioning target mean aortic pressure p for an ex vivo hearts(t) output p from the semi-parametric modelr(t + Deltat) is subtracted to obtain er(t)，er(t) is the input of the virtual PID controller, the output of which obtains the flow ur；
U to output from virtual PID controllerr(t) as input to the semi-parametric model to obtain an output pr(t + Δ t), and p of the isolated heart obtained at this time was measureda(t)；pr(t + Δ t) and pa(t) obtaining an output deviation e of the isolated heart and the semi-parameter model by difference; obtaining control gain through a self-adaptive control algorithm, and controlling the flow rate of perfusion blood by a centrifugal pump through a control gain device;
the control quantity u output by the self-adaptive control algorithm is the rotating speed of the flow speed control unit, and the perfusion flow q is realized by adjusting up(iii) adjustment of (c); the adaptive control algorithm is as follows:
u＝Kee+Krur+Ksps
wherein, Ke、KrAnd KsIs a dimensional coefficient; keAnd KsThe dimension of (A) is rotation speed unit/pressure intensity unit, KrThe dimension of (a) is units of rotation speed/units of flow.
Further, the dimensional coefficient Ke、KrAnd KsThe adaptive update process of (2) is as follows:
Figure BDA0003614387940000041
Figure BDA0003614387940000042
Figure BDA0003614387940000043
Ke1、Ke2、Ks1、Ks2、Kr1、Tcis a constant.
Further, the method comprises the step of storing the obtained data set after adaptive control is carried out for a preset period of time
Figure BDA0003614387940000044
Computing a query point x*Of (d) is a variance sigma (x)*) (ii) a When ∑ (x)*) Greater than a specified threshold sigmamaxIn time, the semi-parametric model of the isolated heart was updated.
The isolated heart perfusion blood pressure self-adaptive control system based on the semi-parametric model comprises a data storage module, a semi-parametric isolated heart model module, a control gain unit and a virtual PID controller;
the data storage module: the device is responsible for recording input values and output values of isolated hearts at any time; the input value is the flow velocity of blood injected into the isolated heart, and the output value is the aortic pressure value of the isolated heart;
the isolated heart semi-parameter model module: establishing a semi-parametric model of an ex vivo heart according to the method of claim 1, 2, 3 or 4;
the virtual PID controller: updating parameters of the virtual PID controller by using a Ziegler-Nichols method; simulating a target average aortic pressure control process according to the input target average aortic pressure and the established isolated heart semi-parameter model;
the control gain unit: determining a control gain by using a self-adaptive control algorithm based on the target average aortic pressure, the actual aortic pressure, the value obtained by the virtual controller and the aortic pressure output by the isolated heart semi-parameter model, and further controlling the blood flow rate injected into the isolated heart;
the adaptive control algorithm is as follows:
u＝Kee+Krur+Ksps
wherein, Ke、KrAnd KsIs a dimensional coefficient; k iseAnd KsThe dimension of (A) is rotation speed unit/pressure intensity unit, KrThe dimension of (a) is a rotation speed unit/flow unit;
having dimensional coefficient Ke、KrAnd KsThe adaptive update process of (2) is as follows:
Figure BDA0003614387940000051
Figure BDA0003614387940000052
Figure BDA0003614387940000053
Ke1、Ke2、Ks1、Ks2、Kr1、Tcis a constant.
Further, the system also comprises a centrifugal pump which executes blood injection of the isolated heart according to the blood flow rate corresponding to the data result of the control gain unit.
Further, the system also comprises a flow probe sensor and a pressure sensor;
the flow probe sensor is used for measuring the flow velocity of blood injected into the isolated heart, and the pressure sensor is used for measuring the aortic pressure value of the isolated heart and storing the obtained numerical values in the storage module.
Has the advantages that:
according to the invention, the in vitro heart parameters are modeled by establishing an in vitro heart reference model. The linear and non-linear properties of an ex vivo heart are well modeled by using a combination of parametric and non-parametric models. Meanwhile, internal parameters of the model can be adaptively changed according to the change of various parameters of the isolated heart by establishing the reference model, and the optimal control gain is generated according to the heart model at the moment, so that the storage time of the isolated heart is prolonged.
According to the invention, a virtual controller is established to generate a reference control track on an isolated heart reference model through the virtual controller. The reference control track can be used as an initial value to better control the aortic pressure of the isolated heart, so that the isolated heart arterial pressure can reach a preset value at a high speed.
The invention establishes a new self-adaptive controller method, which comprehensively considers the target average aortic pressure, the actual average aortic pressure and the reference aortic pressure. By using the target average aortic pressure, the actual average aortic pressure, the isolated heart semi-parameter model and the virtual controller to update the control gain, the convergence speed of the control gain is faster, and a better control effect is achieved.
According to the invention, through establishing the data storage system, the input and the output of the isolated heart at each moment are stored, the stored data are transmitted into the reference model, and the reference model is updated, so that the reference model can be changed in a self-adaptive manner according to the parameter change of the isolated heart.
Therefore, the invention can well solve the problems that in the existing extracorporeal heart perfusion technology, the heart is damaged due to the change of various parameters of an extracorporeal heart model, the blood pressure regulation is difficult to rapidly stabilize, and the heart is damaged when the current controller adjusts the gain due to the changed heart perfusion condition.
Drawings
FIG. 1 is a schematic view of a semi-parametric model;
FIG. 2 is a schematic diagram of a control system;
FIG. 3 is a flow chart of a semi-parametric model update.
Detailed Description
The first embodiment is as follows:
the embodiment is an isolated heart perfusion blood pressure self-adaptive control method based on a semi-parametric model, is a semi-parametric model self-adaptive control method for isolated heart aortic pressure regulation, and is realized on a specific isolated heart perfusion system. The system comprises an isolated heart, a data storage module, a centrifugal pump, a semi-parameter isolated heart model, a control gain unit and a virtual PID controller;
isolated heart: the isolated heart selected in this embodiment is understood to be a human heart or other large mammalian heart that can be used for heart transplantation, and various parameters of the isolated heart may change with time, temperature and other factors. In order not to affect the activity and function of the isolated heart, the isolated heart should maintain a relatively stable aortic pressure value.
A data storage module: the data storage module is mainly responsible for recording input values and output values of isolated hearts at any time. In the embodiment, the flow probe sensor is used for measuring the flow velocity of blood injected into the isolated heart, the pressure sensor is used for measuring the aortic pressure value of the isolated heart, and the obtained numerical values are stored in the storage module.
Isolated heart semi-parametric model: an ex vivo heart semi-reference model was constructed as in figure 1. The parametric model is a ternary model describing the reference relationship between aortic pressure and perfusion flow. The non-parametric model is a data-driven model that describes the non-linearity and uncertainty of the perfusion system. The parametric model and the non-parametric model together form an isolated heart semi-parametric model.
Virtual PID controller: the method used in this embodiment is a Ziegler-Nichols method to update the parameters of the virtual PID controller. And simulating a target average aortic pressure control process according to the input target average aortic pressure and the established isolated heart semi-parameter model, and generating an ideal system track through a virtual controller.
A control gain unit: the self-adaptive algorithm provided by the invention. And comprehensively considering the target average aortic pressure, the actual aortic pressure, the value obtained by the virtual controller and the aortic pressure output by the isolated heart semi-parameter model to obtain a control gain so as to control the flow rate of the blood injected into the isolated heart by the centrifugal pump.
Centrifugal pump: the present embodiment uses a centrifugal pump as a flow rate control unit to control the flow rate of blood input into the aorta. The centrifugal pump chosen should achieve an accurate flow control over the control gain calculated by the gain control unit. The centrifugal pump should have the characteristics of high sensitivity, good control precision and the like.
The control method described in this embodiment specifically includes the steps of:
the method comprises the following steps: according to the attached figure 2, an isolated heart perfusion system is set up, and the input of the isolated heart perfusion system is the target average aortic pressure ps(t) the output is the true aortic pressure p obtained by actual measurementa(t) of (d). The system comprises an ex-vivo heart, a data storage module, a blood flow rate control unit, a control gain unit, an ex-vivo heart reference model, a virtual controller and the like.
Step two: setting a fixed sampling frequency at which the isolated heart is sampled at a flow rate q for perfusing blood into the isolated heartpAortic pressure p of isolated hearta；
Then, a sampling data set is constructed by using the sampling data: the flow rate q of the blood perfused into the isolated heart by the blood flow rate control unit at the current momentpAortic pressure of the isolated heart at the current time and aortic pressure p of the isolated heart at the past D timesaAs input x (t); taking the aortic pressure of the isolated heart perfused by the centrifugal pump at the next sampling moment as output y (t);
x(t)＝[qp(t)，pa(t)，pa(t-Δt)，pa(t-2Δt)…pa(t-DΔt)]∈RD+2
y(t)＝pa(t+Δt)∈R
wherein q ispA flow rate indicative of a flow rate at which the blood flow rate control unit perfuses blood into the isolated heart; p is a radical of formulaaRepresenting the aortic pressure of the isolated heart.Δ t represents a sampling time interval corresponding to the data; d represents the dimension of the input.
The parameter configuration of the parameterized model is recorded as S ═ Rm，Cm，RdAdjusting parameters in S in a perturbation mode to obtain an optimal parameter set S by a minimum Root Mean Square Error (RMSE) between the actual measured aortic pressure and the predicted aortic pressure*：
Figure BDA0003614387940000071
Subsequent Rm、Cm、RdAll adopt S*The arrangement of (1).
Step three: establishing a parameter model in an isolated heart semi-parameter model:
the parametric model is a ternary model. The ternary model uses one capacitance and two resistances to describe the relationship between aortic pressure and perfusion flow. According to kirchhoff's law, the impedance of the aorta is
pa-pm＝Rm·qa
Cm·dpm/dt＝qm
pm＝Rd·(qa-qm)
Wherein p isaShowing the true aortic pressure, q, of the isolated heart as measured by the actual measurementaIs the real aortic flow of the isolated heart obtained by actual measurement; rmIs the flow resistance of blood through the coronary arteries, pmIs RmPressure at the rear node; cmIs the volume of flow in the coronary arteries. q. q.smIs the blood passing through CmThe flow rate of (c). RdIs the flow resistance of blood flowing through the distal coronary arteries.
The parametric model is represented as:
Figure BDA0003614387940000081
Figure BDA0003614387940000082
wherein the content of the first and second substances,
Figure BDA0003614387940000083
a model of a parameter is represented by,
Figure BDA0003614387940000084
representing inputs of a parametric model, qp(t + Δ t) represents the output of the parametric model;
step four: establishing an isolated heart semi-parameter model:
the semi-parametric model is shown in fig. 1, where the non-parametric model part is used to describe the non-linearities and uncertainties involved in the ex vivo heart model. The non-parametric model is a data-driven model, i.e. mapped directly from the inputs to the outputs, rather than relying on some particular equation.
The learning data set is represented as:
Figure BDA0003614387940000085
wherein x isi、yiThe optimal parameter sets x (t), y (t) are in the form of learning data sets; n is the learning data set size.
The input data sampled at a particular frequency follows a gaussian process distribution, so the expression of the non-parametric model f (x) is as follows:
f(x)～GP(μ(x)，k(x，x′))
wherein GP (·) represents a Gaussian distribution; μ (x) represents the mean function of the input data x (t), k (x, x ') represents the covariance function of any two input data x and x';
μ(x)＝E[f(x)]
k(x，x′)＝E[(f(x)-μ(x))(f(x′)-μ(x′))]
wherein E [. cndot. ] represents a mathematical expectation; using parametric models as a priori knowledge, i.e.
Figure BDA0003614387940000086
Figure BDA0003614387940000087
Wherein
Figure BDA0003614387940000088
Thus, the semi-parametric model is expressed as follows:
f(x)～GP(G(xA)，k(x，x′))
let x*The output of the semi-parametric model is the predicted average f (x) of the results for the query point of the model*) Sum variance Σ (x)*) (ii) a For query point x*The predicted results are as follows:
Figure BDA0003614387940000089
Figure BDA00036143879400000810
wherein
y＝[y1，…，yN]T
Figure BDA0003614387940000091
μ(X)＝[μ(x1)，…，μ(xN)]T
K＝k(X，X)
Km，n＝k(xm，xn)
k*＝k(X，x*)
km，*＝k(xm，x*)
k**＝k(x*，x*)
Wherein σnIs x*I is a unit matrix;
based on the above, a semi-parametric model is obtained, and in the block diagram of fig. 2, the input of the semi-parametric model is x (t) ═ ur(t)，pr(t)，pr(t-Δt)，pr(t-2Δt)…pr(t-DΔt)]The output is y (t) ═ pr(t+Δt)。
Step five: based on the obtained semi-parameter model, testing the semi-parameter model by utilizing an open-loop step response mode in a Ziegler-Nichols method, and then obtaining parameters of the virtual PID controller by utilizing an overshoot-free method of the Ziegler-Nichols method;
maintaining vital and optimally functioning target mean aortic pressure p for an ex vivo hearts(t) output p from the semi-parametric modelr(t + Deltat) is subtracted to obtain er(t)，er(t) is the input of the virtual PID controller, the output of which obtains the flow ur。
Step six: u to output from virtual PID controllerr(t) As input to the semi-parametric model, an output p can be obtainedr(t + Δ t), and p of the isolated heart obtained at this time was measureda(t)；pr(t + Δ t) and pa(t) obtaining an output deviation e of the isolated heart and the semi-parameter model by difference; obtaining control gain through a self-adaptive control algorithm, and controlling the flow rate of the perfusion blood by a centrifugal pump through a control gain controller.
The invention provides a new self-adaptive control algorithm for adjusting the control gain, and the self-adaptive control algorithm adopts the variable to adjust the control gain. The control quantity u output by the control algorithm is the rotating speed of the flow speed control unit, and the perfusion flow q is realized by adjusting upAnd (4) adjusting. The control algorithm is as follows:
u＝Kee+Krur+Ksps
wherein Ke、KrAnd KsAre dimensional coefficients. KeAnd KsIs the rotation speedUnit/pressure Unit (/ denotes "ratio"), KrThe dimension of (d) is rotational speed unit/flow unit. The self-adaptive updating algorithm comprises the following steps:
Figure BDA0003614387940000092
Figure BDA0003614387940000093
Figure BDA0003614387940000101
Ke1、Ke2、Ks1、Ks2、Kr1、Tcis constant and is adjusted according to actual conditions.
Step seven: after a predetermined period of time, the data set obtained by storing
Figure BDA0003614387940000102
Computing a query point x*Of (d) is a variance sigma (x)*). When ∑ (x)*) And returning to the step three when the value is larger than the specified threshold value sigma max. The updating process is as shown in figure 3.
The present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof, and it is therefore intended that all such changes and modifications be considered as within the spirit and scope of the appended claims.

Claims (10)

1. A method of modeling an ex vivo heart, comprising the steps of:
firstly, establishing a parameter model in an isolated heart semi-parameter model, wherein the parameter model is a ternary model, the ternary model describes the relation between the aortic pressure and the perfusion flow by using a capacitor and two resistors, and the resistor R ismRear parallel connection resistor RdAnd a capacitor Cm，RmIs the flow resistance of blood through the coronary arteries, CmIs the volume of flow of the coronary artery, RdIs the flow resistance of blood flowing through the distal coronary artery;
then determining the impedance of the aorta according to kirchhoff's law based on the ternary model;
the parametric model is represented as:
Figure FDA0003614387930000011
Figure FDA0003614387930000012
wherein the content of the first and second substances,
Figure FDA0003614387930000013
a model of a parameter is represented by,
Figure FDA0003614387930000014
representing inputs of a parametric model, qp(t + Δ t) represents the output of the parametric model;
describing nonlinearities and uncertainties contained in the ex vivo heart model using a non-parametric model; the learning data set is represented as:
Figure FDA0003614387930000015
wherein x isi、yiInput data x (t) and output data y (t) in the learning data set; n is the learning dataset size;
x(t)＝[qp(t)，pa(t)，pa(t-Δt)，pa(t-2Δt)...pa(t-DΔt)]∈RD+2
y(t)＝pa(t+Δt)∈R
wherein q ispMeans for controlling blood flow rate to perfuse blood into isolated heartThe flow rate of (a); p is a radical ofaThe aortic pressure of the isolated heart is represented, and delta t represents the sampling time interval corresponding to the data; d represents the dimension of the input;
the sampled input data follows a Gaussian process distribution, and a non-parametric model f (x) is expressed as follows:
f(x)～GP(μ(x)，k(x，x′))
wherein GP (·) represents a Gaussian distribution; μ (x) represents the mean function of the input data x (t), k (x, x ') represents the covariance function of any two input data x and x';
μ(x)＝E[f(x)]
k(x，x′)＝E[(f(x)-μ(x))(f(x′)-μ(x′))]
wherein E [. cndot. ] represents a mathematical expectation; using parametric models as a priori knowledge, i.e.
Figure FDA0003614387930000016
Figure FDA0003614387930000017
Wherein
Figure FDA0003614387930000018
The semi-parameter model of the isolated heart is f (x) -GP (G (xA), k (x, x')).
2. The ex vivo cardiac modeling method of claim 1, wherein the aortic impedance determined according to kirchhoff's law based on the trigram is as follows:
pa-pm＝Rm·qa
Cm·dpm/dt＝qm
pm＝Rd·(qa-qm)
wherein p isaShowing the actual measured isolated heart aortic pressure, qaIs the real aortic flow of the isolated heart obtained by actual measurement; p is a radical ofmIs RmPressure at the rear node; q. q.smIs the blood passing through CmThe flow rate of (c).
3. An ex vivo cardiac modelling method according to claim 2, wherein the input data x (t) and the output data y (t) of the learning data set are data of an optimal parameter set determined from the sampled data, the process of determining the optimal parameter set from the sampled data comprising the steps of:
firstly, acquiring sampling data, wherein the sampling data is the flow rate q for perfusing blood into an isolated heartpAortic pressure p of isolated hearta；
Then, a sampling data set is constructed by using the sampling data: the flow rate q of the blood perfused into the isolated heart by the blood flow rate control unit at the current momentpAortic pressure of the isolated heart at the current time and aortic pressure p of the isolated heart at the past D timesaAs input x (t); taking the aortic pressure of the isolated heart perfused by the centrifugal pump at the next sampling moment as output y (t);
based on the sampled data set, the parameter configuration of the parameterized model is recorded as S ═ Rm，Cm，RdAdjusting parameters in S in a perturbation mode to enable the minimum root mean square error RMSE between the actual measured aortic pressure and the predicted aortic pressure to obtain an optimal parameter set S*。
4. An ex vivo cardiac modeling method as claimed in claim 3, characterized by querying point x for the model based on a semi-parametric model of the ex vivo heart*The predicted results are as follows:
Figure FDA0003614387930000021
Figure FDA0003614387930000022
wherein
y＝[y1，…，yN]T
Figure FDA0003614387930000023
μ(X)＝[μ(x1)，…，μ(xN)]T
K＝k(X，X)
Km，n＝k(xm，xn)
k*＝k(X，x*)
km，*＝k(xm，x*)
k**＝k(x*，x*)
Wherein σnIs x*I is a unit matrix.
5. The self-adaptive control method for the perfusion blood pressure of the isolated heart based on the semi-parameterized model is characterized by comprising the following steps of:
the semi-parametric model of the isolated heart established based on the modeling method of the isolated heart of claim 1, 2, 3 or 4, the semi-parametric model is tested by using an open-loop step response mode in a Ziegler-Nichols method, and then parameters of the virtual PID controller are obtained by using an overshoot-free method of the Ziegler-Nichols method;
maintaining vital and optimally functioning target mean aortic pressure p for an ex vivo hearts(t) output p from the semi-parametric modelr(t + Δ t) is subtracted to obtain er(t)，er(t) is the input of the virtual PID controller, the output of which obtains the flow ur；
Transmitting the virtual PID controllerU is outr(t) as input to the semi-parametric model to obtain an output pr(t + Δ t), and p of the isolated heart obtained at this time was measureda(t)；pr(t + Δ t) and pa(t) obtaining an output deviation e of the isolated heart and the semi-parameter model by difference; obtaining control gain through a self-adaptive control algorithm, and controlling the flow rate of perfusion blood by a centrifugal pump through a control gain device;
the control quantity u output by the self-adaptive control algorithm is the rotating speed of the flow speed control unit, and the perfusion flow q is realized by adjusting up(iii) adjustment of (c); the adaptive control algorithm is as follows:
u＝Kee+Krur+Ksps
wherein, Ke、KrAnd KsIs a dimensional coefficient; keAnd KsThe dimension of (A) is rotation speed unit/pressure intensity unit, KrThe dimension of (a) is units of rotation speed/units of flow.
6. The semi-parameterized model-based adaptive control method for ex vivo cardiac perfusion blood pressure according to claim 5, wherein the dimensional coefficient K ise、KrAnd KsThe adaptive update process of (2) is as follows:
Figure FDA0003614387930000031
Figure FDA0003614387930000032
Figure FDA0003614387930000033
Ke1、Ke2、Ks1、Ks2、Kr1、Tcis a constant.
7. The semi-parameterized model-based ex vivo cardiac perfusion blood pressure adaptive control method of claim 6, which adaptively controls the blood pressure by storing the obtained data set after a preset period of time
Figure FDA0003614387930000034
Computing a query point x*Of (d) is a variance sigma (x)*) (ii) a When ∑ (x)*) Greater than a specified threshold sigmamaxIn time, the semi-parametric model of the isolated heart was updated.
8. The self-adaptive control system for the perfusion blood pressure of the isolated heart based on the semi-parametric model is characterized by comprising a data storage module, a semi-parametric isolated heart model module, a control gain unit and a virtual PID controller;
the data storage module: the device is responsible for recording input values and output values of isolated hearts at any time; the input value is the flow velocity of blood injected into the isolated heart, and the output value is the aortic pressure value of the isolated heart;
the isolated heart semi-parameter model module: establishing a semi-parametric model of an ex vivo heart according to the method of claim 1, 2, 3 or 4;
the virtual PID controller: updating parameters of the virtual PID controller by using a Ziegler-Nichols method; simulating a target mean aortic pressure control process according to the input target mean aortic pressure and the established isolated heart semi-parameter model;
the control gain unit: determining a control gain by using a self-adaptive control algorithm based on the target average aortic pressure, the actual aortic pressure, the value obtained by the virtual controller and the aortic pressure output by the isolated heart semi-parameter model, and further controlling the blood flow rate injected into the isolated heart;
the adaptive control algorithm is as follows:
u＝Kee+Krur+Ksps
wherein, Ke、KrAnd KsIs a dimensional coefficient;Keand KsThe dimension of (a) is unit of rotation speed/unit of pressure, KrThe dimension of (a) is a rotation speed unit/flow unit;
having dimensional coefficient Ke、KrAnd KsThe adaptive update process of (2) is as follows:
Figure FDA0003614387930000041
Figure FDA0003614387930000042
Figure FDA0003614387930000043
Ke1、Ke2、Ks1、Ks2、Kr1、Tcis a constant.
9. The semi-parameterized model-based isolated heart perfusion blood pressure adaptive control system according to claim 8, further comprising a centrifugal pump that performs isolated heart blood injection at a blood flow rate corresponding to the data result of the control gain unit.
10. The semi-parameterized model-based ex vivo cardiac perfusion blood pressure adaptive control system of claim 9, further comprising a flow probe sensor and a pressure sensor;
the flow probe sensor is used for measuring the flow velocity of blood injected into the isolated heart, the pressure sensor is used for measuring the aortic pressure value of the isolated heart, and all obtained numerical values are stored in the storage module.
CN202210439215.8A 2022-04-25 2022-04-25 Isolated heart perfusion blood pressure self-adaptive control method based on semi-parametric model Active CN114740728B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210439215.8A CN114740728B (en) 2022-04-25 2022-04-25 Isolated heart perfusion blood pressure self-adaptive control method based on semi-parametric model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210439215.8A CN114740728B (en) 2022-04-25 2022-04-25 Isolated heart perfusion blood pressure self-adaptive control method based on semi-parametric model

Publications (2)

Publication Number Publication Date
CN114740728A true CN114740728A (en) 2022-07-12
CN114740728B CN114740728B (en) 2023-03-03

Family

ID=82283816

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210439215.8A Active CN114740728B (en) 2022-04-25 2022-04-25 Isolated heart perfusion blood pressure self-adaptive control method based on semi-parametric model

Country Status (1)

Country Link
CN (1) CN114740728B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116235846A (en) * 2023-02-27 2023-06-09 中国科学院苏州生物医学工程技术研究所 Organ perfusion control method, system and device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1211930A (en) * 1996-01-08 1999-03-24 新技术(Sa－Ysy)有限公司 Electrical muscle controller
CN200979542Y (en) * 2006-07-31 2007-11-21 胡晓旻 A novel perfusion device for isolated hearts
CN111407244A (en) * 2020-03-30 2020-07-14 上海大学 Left ventricle end-systolic elasticity prediction method and system based on SVR algorithm
EA202000057A1 (en) * 2019-12-26 2021-06-30 Дмитрий Алексеевич Федорук DEVICE FOR PERFUSION CONDITIONING OF DONOR BODY

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1211930A (en) * 1996-01-08 1999-03-24 新技术(Sa－Ysy)有限公司 Electrical muscle controller
CN200979542Y (en) * 2006-07-31 2007-11-21 胡晓旻 A novel perfusion device for isolated hearts
EA202000057A1 (en) * 2019-12-26 2021-06-30 Дмитрий Алексеевич Федорук DEVICE FOR PERFUSION CONDITIONING OF DONOR BODY
CN111407244A (en) * 2020-03-30 2020-07-14 上海大学 Left ventricle end-systolic elasticity prediction method and system based on SVR algorithm

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
LIMING XIN ET AL.: "A New Multi-Mode Perfusion System for Ex Vivo Heart Perfusion Study", 《SPRINGER》 *
LIMING XIN ET AL.: "Model Reference Adaptive Control for Aortic Pressure Regulation in Ex Vivo Heart Perfusion", 《IEEE》 *
LIMING XIN ET AL.: "Primed Left Ventricle Heart Perfusion Creates Physiological Aortic Pressure in Porcine Hearts", 《ASAIO JOURNAL 2020》 *
李霞等: "心脏超声建模的制作体会", 《中国医药创新》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116235846A (en) * 2023-02-27 2023-06-09 中国科学院苏州生物医学工程技术研究所 Organ perfusion control method, system and device
CN116235846B (en) * 2023-02-27 2024-01-30 中国科学院苏州生物医学工程技术研究所 Organ perfusion control method, system and device

Also Published As

Publication number Publication date
CN114740728B (en) 2023-03-03

Similar Documents

Publication Publication Date Title
CN114740728B (en) Isolated heart perfusion blood pressure self-adaptive control method based on semi-parametric model
CN108284442B (en) Mechanical arm flexible joint control method based on fuzzy neural network
Choi et al. A sensorless approach to control of a turbodynamic left ventricular assist system
CN111529047A (en) Cryoablation temperature control method, system and medium
KR100331093B1 (en) Devices for in-vivo determination of the compliance function and the systemic blood flow of a living being
Petrou et al. A versatile hybrid mock circulation for hydraulic investigations of active and passive cardiovascular implants
Misgeld et al. Robust decentralised control of a hydrodynamic human circulatory system simulator
Burattini et al. Complex and frequency-dependent compliance of viscoelastic windkessel resolves contradictions in elastic windkessels
Stephens et al. In vitro evaluation of an immediate response Starling‐like controller for dual rotary blood pumps
Fetanat et al. A sensorless control system for an implantable heart pump using a real-time deep convolutional neural network
Stern et al. The self-tuning controller: Comparison with human performance in the control of arterial pressure
Li et al. Fuzzy-PID control for the regulation of blood glucose in diabetes
CN116173397A (en) Rotational speed control device and method, ventricular assist system, storage medium, and apparatus
Pigot et al. The differential-algebraic Windkessel model with power as input
Silva et al. A variable gain physiological controller for a rotary left ventricular assist device
Gwak et al. Feedback linearization control of a cardiovascular circulatory simulator
Xin et al. Model reference adaptive control for aortic pressure regulation in ex vivo heart perfusion
Shi et al. Mathematical and physical models of a total artificial heart
Telyshev et al. Hybrid mock circulatory loop for training and study purposes
van Gerven et al. An elastance-based control scheme for a heart valve tester
Sims Estimation of arterial system parameters from dynamic records
Timmons Cardiovascular models and control
Fisher et al. A controllable artificial afterload for isolated heart studies
CN116235846B (en) Organ perfusion control method, system and device
Quick et al. The arterial system pressure–volume loop

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant