Self-matching control method of in-vitro simulation circulation system
By collecting signals in real time in the in vitro simulation circulation system and adjusting parameters, designing status and reward function using numerical models and solenoid proportional valves, the problem that arterial pressure and flow rate of patients with central aging in vitro simulation circulation system is not in line with the physiological characteristics of the human body, and the problem is realized Fast and high-precision multi-objective control.
Patent Information
- Application Number
- CN202410912147.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-08
- Publication Date
- 2024-11-08
AI Technical Summary
The prior art is difficult to accurately simulate the arterial pressure and flow characteristics of heart failure patients in in vitro simulation circulatory systems, and the multi-objective control pattern is complex, making it difficult to achieve effective self-match control.
Signals are collected in real time through multiple sensors, target heart rate, target arterial systolic pressure and target average arterial flow, numerical model and solenoid proportional valve to adjust systolic pressure, diastolic pressure, arterial compliance and resistance, design status and reward function, and gradually adjust it Parameters to achieve the target value.
It realizes rapid, high-precision, and large-scale simulation of arterial pressure and flow characteristics of patients with heart failure, improves the range and accuracy of the simulation, effectively anti-interference, and ensures the proximity of multiple target values.
Smart Images

Figure CN118915434A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of medical devices, in particular to a self-matching control method for an in vitro simulated circulation system. Background Art
[0002] The physiological characteristics of arterial pressure and flow are important indicators for cardiovascular patients. Changes in arterial pressure and flow reflect many diseases. Accurate simulation of arterial physiological characteristics in vitro over a large range can better study the pathological characteristics of cardiovascular patients, and can also provide a more extensive and accurate environment for the development and control of cardiovascular equipment. Simulating the arterial pressure and flow of specific heart failure patients by adjusting multiple parameters in an in vitro simulated circulation system is a multi-input and multi-output nonlinear system control problem. It is difficult to establish a control model for the actual system and the control law of multiple objectives is often very complicated. A self-matching control method for an in vitro simulated circulation system is proposed. A large number of simulation trainings are carried out on the designed self-matching control method for an in vitro simulated circulation system, which can simulate the arterial pressure and flow characteristics of heart failure patients in accordance with the physiological state of the human body, and has good anti-interference ability. Summary of the invention
[0003] The purpose of the present invention is to provide a self-matching control method for an extracorporeal simulated circulatory system, which can solve the problem that the arterial pressure and flow of simulated heart failure patients do not conform to the physiological characteristics of the human body.
[0004] To achieve the above object, the technical solution of the present invention is: a self-matching control method of an in vitro simulated circulatory system, characterized in that it is implemented according to the following steps:
[0005] S1: Use multiple sensors to collect feedback signals from the in vitro simulated circulatory system in real time;
[0006] S2: Set the target heart rate HR within the control range according to the physiological characteristics of heart failure patients t , target arterial systolic pressure Target diastolic blood pressure and target mean arterial flow
[0007] S3: Initial generation parameter contraction pressure P e , diastolic pressure P f , arterial compliance C s , the resistance R simulated by the electromagnetic proportional valve s The input system numerical model runs for 10 cardiac cycles. The numerical model needs a certain amount of time to reach a steady state. The average value of the last 5 cardiac cycles is taken as the measurement value.
[0008] S4: The initial state s is obtained by subtracting the real-time measured value of the in vitro simulated circulation system from the target value t=[ΔAP sys , ΔAP dia , ΔAQ mean ];
[0009] S5: Based on the sum of squared deviations S 2 Determine the current state t Output the action a to be performed next t , S 2 If it does not exceed 4, it means that the absolute value of the deviation between all measured values and the target value is within 2, and the output action is stopped;
[0010] S6: If S 2 Exceeding 4, according to the initial state s t Output the action a to be performed next t , and the action a t After processing, the input gets a new state s t+1 and reward r t , according to the new state s t+1 Output the action a to be performed next t+1 , and keep adjusting these four parameters until S 2 Not more than 4;
[0011] S7: When the control of the in vitro simulated circulation system reaches the target value of regulation, the regulation ends automatically.
[0012] In one embodiment of the present invention, in step S6, the partial reward function is defined as
[0013]
[0014]
[0015] Multi-target control is about ensuring that multiple measurements under regulation approach multiple targets simultaneously.
[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention proposes a self-matching control method for an in vitro simulated circulatory system, which can simulate the arterial pressure and flow characteristics of specific heart failure patients quickly, with high precision and over a large range. Kirchhoff's law derives the state equation, designs the state, action and reward functions, and effectively improves the range and accuracy of in vitro simulation of arterial characteristics of specific heart failure patients. A self-matching control method for an in vitro simulated circulatory system is proposed to solve the problem that the arterial pressure and flow of simulated heart failure patients do not conform to the physiological characteristics of the human body. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 The present invention is a flow chart of a control method for an in vitro simulated circulatory system.
[0018] Figure 2is the sum of squared deviations in the control of the in vitro simulated circulatory system 2 Diagram of the regulation process.
[0019] Figure 3 Diagram of the arterial pressure regulation process in the control of the in vitro simulated circulatory system.
[0020] Figure 4 Diagram of the average flow control process of the control artery in the in vitro simulated circulatory system. DETAILED DESCRIPTION
[0021] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings:
[0022] The present invention provides a control method for an in vitro simulated circulatory system, characterized in that: Figure 1 As shown, follow the steps below to implement it:
[0023] S1: Use multiple sensors to collect feedback signals from the in vitro simulated circulatory system in real time;
[0024] S2: Set the target heart rate HR within the control range according to the physiological characteristics of heart failure patients t , target arterial systolic pressure Target diastolic blood pressure and target mean arterial flow
[0025] S3: Initial generation parameter contraction pressure P e , diastolic pressure P f , arterial compliance C s , the resistance R simulated by the electromagnetic proportional valve s The input system numerical model runs for 10 cardiac cycles. The numerical model needs a certain amount of time to reach a steady state. The average value of the last 5 cardiac cycles is taken as the measurement value.
[0026] S4: The initial state s is obtained by subtracting the real-time measured value of the in vitro simulated circulation system from the target value t =[ΔAP sys , ΔAP dia , ΔAQ mean ];
[0027] S5: Based on the sum of squared deviations S 2 Determine the current state t Output the action a to be performed next t , S 2 If it does not exceed 4, it means that the absolute value of the deviation between all measured values and the target value is within 2, and the output action is stopped;
[0028] S6: If S 2 Exceeding 4, according to the initial state st Output the action a to be performed next t , and the action a t After processing, the input gets a new state s t+1 and reward r t , according to the new state s t+1 Output the action a to be performed next t+1 , and keep adjusting these four parameters until S 2 Not exceeding 4.
[0029] S7: When the control of the in vitro simulated circulation system reaches the target value of regulation, the regulation ends automatically.
[0030] Furthermore, in step S2, the target heart rate HR t After entering the numerical model, it remains unchanged and does not need to be adjusted. AQ mean ∈[20,50]×10 -1 L / min, the simulation target value is: HR t =75bpm / min, AQ mean =28×10 -1 L / min.
[0031] Furthermore, in step S2, the state is designed in order to learn the control strategy under different target values, and the difference between the system measurement value and the patient's target value is selected as the state. How to output actions under different deviations to obtain greater rewards. State s t is defined as
[0032]
[0033]
[0034] In the formula Since it takes a certain amount of time to reach a steady state, the average value of the last five cardiac cycles is selected as the measurement value. AP(t) and AQ(t) are real-time measurement values of the system. and is the target value, △AP sys and ΔAP dia The unit is mmHg. In order to improve the flow control accuracy, the flow value is enlarged to the same level as the pressure, so △AQ mean The unit is 10 -1 L / min.
[0035] Furthermore, in step S3, the action space is designed. The arterial pressure and flow of the human body are mainly affected by the systolic and diastolic capacity of the heart, arterial compliance, vascular resistance and heart rate. The systolic and diastolic capacity of the left ventricle is mainly affected by the compressed gas pressure P e and vacuum pressure P f The arterial compliance and resistance of the resistance simulator are respectively determined by the lumped parameter C s and R s Indicates that the heart rate can convert the contraction and relaxation signal P o Control. Select the change of each parameter as the output action, action a t Defined as
[0036] a t =[△P e △P f △C s △R s ]
[0037] The output action range is from -1 to 1, so you need to t Scaling is performed according to a certain ratio, and the actual output parameter change is [100×△P e mmHg, 10×△P f mmHg, 0.6×△C s ml / mmHg, 0.6×△R s mmHg·s / ml], and then add the change to the original value to get the new value.
[0038] Furthermore, in step S6, the reward function is designed to judge the current strategy. When it is close to the target value, a positive reward should be given, and when it is far away from the target, a negative reward should be given. A suitable reward function can guide the process in the right direction, reduce trial and error, and speed up the training process. Unlike a single target value, the reward function design for multiple target values requires learning to approach multiple targets at the same time: the measured value is very close to one or two target values, but has a large deviation from other target values. In order to prevent a large reward from being caused by approaching a single target, the reward function is defined as
[0039]
[0040] The key point of multi-target control is to ensure that multiple measured values in regulation are close to multiple targets at the same time. Therefore, the square root of the deviation between each measured value and the target value is selected as an item in the reward function, which can avoid bringing a large reward when approaching a single target. Only when all three measured values are close to the target value, the square root of the deviation between the measured value and the target value will decrease, and then a large reward can be obtained. Figure 2 , Figure 3 , Figure 4As shown, it means that the absolute value of the deviation between all measured values and the target value is within 2. When the absolute value of the deviation between the three measured values and the target value does not exceed 4, a positive reward is given, and the upper limit of a single reward is 30. When the absolute value of any deviation is outside 4, a negative value is given. When the deviation is very large, the single reward or punishment tends to -4. When the deviation becomes smaller, the slope of the reward becomes larger, speeding up the training.
Claims
1. A self-matching control method for an in vitro simulated circulatory system, It is characterized in that Follow these steps to achieve this: S1: Use multiple sensors to collect feedback signals from the in vitro simulated circulatory system in real time; S2: Set the target heart rate HR within the control range according to the physiological characteristics of heart failure patients t , target arterial systolic pressure Target diastolic blood pressure and target mean arterial flow S3: Initial generation parameter contraction pressure P e , diastolic pressure P f , arterial compliance C s , the resistance R simulated by the electromagnetic proportional valve s The input system numerical model runs for 10 cardiac cycles. The numerical model needs a certain amount of time to reach a steady state. The average value of the last 5 cardiac cycles is taken as the measurement value. S4: The initial state s is obtained by subtracting the real-time measured value of the in vitro simulated circulation system from the target value t =[△AP sys ,△AP dia ,△AQ mean ]; S5: Based on the sum of squared deviations S 2 Determine the current state t Output the action a to be performed next t , S 2 If it does not exceed 4, it means that the absolute value of the deviation between all measured values and the target value is within 2, and the output action is stopped; S6: If S 2 Exceeding 4, according to the initial state s t Output the action a to be performed next t , and the action a t After processing, the input gets a new state s t+1 and reward r t , according to the new state s t+1 Output the action a to be performed next t+1 , and keep adjusting these four parameters until S 2 No more than 4; S7: When the control of the in vitro simulated circulation system reaches the target value of regulation, the regulation ends automatically.
2. The self-matching control method of an in vitro simulated circulatory system according to claim 1, Features: In step S2, the target heart rate HR t After entering the in vitro simulated circulatory system, it remains unchanged and does not require regulation. AQ mean ∈[20,50]×10 -1 L / min, the simulation target value is: HR t =75bpm / min, AQ mean =28×10 -1 L / min.
3. The self-matching control method of an in vitro simulated circulatory system according to claim 1, Features: In step S4, how to output actions under deviation to obtain greater rewards, state s t is defined as In the formula Since the numerical model needs a certain amount of time to reach a steady state, the average value of the last five cardiac cycles was selected as the measurement value.
4. The self-matching control method of an in vitro simulated circulatory system according to claim 1, Features: In step S6, the absolute values of the deviations between the measured values and the target values are all within 2. When the absolute values of the deviations between the three measured values and the target values do not exceed 4, a positive reward is given, and the upper limit of a single reward is 30. When the absolute value of any deviation is outside 4, a negative value is given. When the deviation is very large, the single reward or punishment tends to -4. When the deviation becomes smaller, the slope of the reward becomes larger.
5. The self-matching control method of an in vitro simulated circulatory system according to claim 1, Features: In step S6, in order to prevent a large reward from being caused by approaching a single target, the reward function is defined as Multi-target control ensures that multiple measured values in regulation approach multiple targets at the same time. Therefore, the square root of the deviation between each measured value and the target value is selected as an item in the reward function. This can avoid bringing a large reward when approaching a single target. Only when all three measured values are close to the target value will the square root of the deviation between the measured value and the target value decrease, and only then can a larger reward be obtained.
Citation Information
Cited By
Ultrasonic actuating pump lung control method based on reinforcement learning and fuzzy control
CN119846956A
