An adaptive PID control method and device for a blood pressure simulator

Abstract

The application relates to the technical field of medical instrument detection, and discloses a self-adaptive PID control method and device for a blood pressure simulator, which comprises the following steps: online identification of approximate model parameters of a gas pressure system under a current working condition; real-time calculation and updating of parameters of a PID controller by using an intelligent reasoning mechanism based on the obtained approximate model parameters; and closed-loop control of the gas pressure system by using the updated PID controller parameters. The device comprises a main control module, an online identification module and an intelligent parameter setting module to execute the method. The application can further comprise performance evaluation and self-learning steps, and a self-learning process is triggered to fine-tune the control strategy when the control performance exceeds a preset range. The application realizes high-precision and strong self-adaptive gas pressure control, and improves the intelligent level and robustness of the device.

Description

Technical Field

[0001] This application relates to the field of medical device testing technology, and in particular to an adaptive PID control method and device for a blood pressure simulator. Background Technology

[0002] Blood pressure simulators are key metrological instruments used to test and calibrate the performance of medical devices such as electronic blood pressure monitors. One of their core functions is to accurately generate and stably control the pressure within the pneumatic system.

[0003] In existing technologies, the pneumatic control system of blood pressure simulators generally employs a proportional-integral-derivative (PI-DE) control algorithm. However, this traditional control method typically has inherent drawbacks. First, its core control parameters, such as the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd, are usually tuned to a fixed set of values ​​at the factory. However, the actual operating environment of a blood pressure simulator is variable; for example, it needs to operate over a wide pressure range or in different modes such as continuous deflation or stepped deflation. The dynamic characteristics of the controlled object (including the pneumatic path and cuff load) change with variations in pressure, load, and other operating conditions. Fixed control parameters cannot maintain optimal control performance under all operating conditions, resulting in poor system adaptability and a difficulty in simultaneously achieving control accuracy and response speed.

[0004] To address this issue, some improved methods have introduced adaptive concepts, such as adaptively adjusting a coefficient in the feedforward control based on pressure error. However, these methods do not fundamentally solve the core problem of adaptive proportional-integral-derivative (PID) control parameters. When the system model itself changes due to factors such as component aging or minor gas leaks, the control performance still deteriorates. Furthermore, the traditional control parameter tuning process heavily relies on the experience of engineers, which is not only time-consuming and labor-intensive, but also prevents the equipment from self-adjusting and compensating for performance degradation caused by environmental changes or its own aging after it has been put into use.

[0005] Therefore, how to achieve high-precision and highly adaptable air pressure control under all operating conditions by sensing changes in the dynamic characteristics of the system in real time and automatically adjusting the core control parameters is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] This application provides an adaptive PID control method and apparatus for a blood pressure simulator, aiming to solve the problems of fixed controller parameters, poor adaptability, inability to adapt to changes in system dynamic characteristics online, and complex debugging in the prior art.

[0007] According to a first aspect of the present disclosure, this application provides an adaptive PID control method for a blood pressure simulator, employing the following technical solution:

[0008] An adaptive PID control method for a blood pressure simulator, comprising:

[0009] Online identification steps: Identify the approximate model parameters of the pneumatic system of the blood pressure simulator under the current operating conditions online;

[0010] Intelligent parameter tuning steps: Based on the approximate model parameters obtained in the online identification step, the PID controller parameters are calculated and updated in real time using an intelligent inference mechanism; and

[0011] Adaptive control step: Use the updated PID controller parameters from the intelligent parameter tuning step to perform closed-loop control of the air pressure system.

[0012] Optional, also includes:

[0013] Performance evaluation and self-learning steps: Periodically evaluate the performance indicators of the control process, and when the performance indicators exceed the preset performance range, trigger the self-learning process to fine-tune the preset control strategy on which the intelligent reasoning mechanism is based.

[0014] Optionally, the preset control strategy based on the intelligent reasoning mechanism is an expert rule base, and the self-learning process includes fine-tuning the expert rule base.

[0015] Optionally, the online identification step specifically includes:

[0016] A preset excitation signal is superimposed on the control quantity of the actuator drive module; and

[0017] Based on the excitation signal and pressure output data, the approximate model parameters are identified using the recursive least squares method.

[0018] Optionally, the approximate model is a first-order inertial plus pure time-delay model:

[0019] ;

[0020] Where K represents the system gain, reflecting the degree of influence of the control quantity change on the steady-state pressure value; T represents the system time constant, reflecting the speed of the system response; and τ represents the pure time delay, reflecting the delay from the issuance of the control quantity to the start of the system response.

[0021] Optionally, the intelligent reasoning mechanism is a fuzzy reasoning mechanism.

[0022] According to a second aspect of the present disclosure, this application provides an adaptive PID control device for a blood pressure simulator, employing the following technical solution:

[0023] An adaptive PID control device for a blood pressure simulator includes a main control module, a pressure sensing module, and an actuator drive module, and further includes:

[0024] The online identification module is configured to identify the approximate model parameters of the air pressure system of the blood pressure simulator online;

[0025] The intelligent parameter tuning module is configured to calculate and update the PID controller parameters in real time based on the approximate model parameters using an intelligent inference mechanism.

[0026] The main control module includes a PID controller, which is configured to set the PID controller using the updated PID controller parameters from the intelligent parameter tuning module, and then perform closed-loop control of the pneumatic system through the PID controller and the actuator drive module.

[0027] Optional, also includes:

[0028] The performance evaluation and self-learning module is configured to periodically evaluate the performance indicators of the control process, and when the performance indicators exceed the preset range, trigger the self-learning process to fine-tune the preset control strategy on which the intelligent parameter tuning module is based.

[0029] Optionally, the intelligent parameter tuning module is a fuzzy inference engine, including a fuzzification interface, a knowledge base (containing a database and a rule base), an inference engine, and a defuzzification interface. Its inputs are the normalized error and the error change rate, and its output is the adjustment amount of the PID parameters.

[0030] Compared with existing technologies, the technical solution provided in this application has the following beneficial effects: 1. High precision and strong adaptability. By online identification and real-time acquisition of the system model, and combined with an intelligent inference mechanism to tune the core parameters of the PID controller online, the controller can automatically adapt to the dynamic characteristics changes of the controlled object under different operating conditions, solving the problems of fixed parameters and poor adaptability of traditional controllers, thus achieving high-precision air pressure control over a wide pressure range and under different loads. 2. Intelligent and maintenance-free. By introducing performance evaluation and self-learning mechanisms, the device can periodically evaluate control performance and self-optimize the control strategy to compensate for performance degradation caused by component aging, environmental changes, etc., significantly reducing reliance on manual experience for debugging and subsequent maintenance costs, and improving the intelligence level of the device. 3. Excellent dynamic performance and robustness. For pressure change conditions such as stepped venting, this application can achieve a fast and overshoot-free response through rapid parameter adaptation, significantly improving the dynamic response capability of the device; at the same time, real-time mastery of the system model and intelligent control strategy can effectively suppress internal and external interferences such as gas path leakage, ensuring the stability and reliability of the control process.

[0031] Other features and advantages disclosed in this invention will be described in detail in the following detailed description section. Attached Figure Description

[0032] The accompanying drawings are provided to further illustrate the present disclosure and form part of the specification. They are used together with the following detailed description to explain the present disclosure, but do not constitute a limitation thereof. In the drawings:

[0033] Figure 1 This is an overall block diagram illustrating an adaptive PID control device for a blood pressure simulator according to an exemplary embodiment.

[0034] Figure 2 This is a flowchart (a) illustrating an adaptive PID control method for a blood pressure simulator according to an exemplary embodiment.

[0035] Figure 3 This is a flowchart (II) illustrating an adaptive PID control method for a blood pressure simulator according to an exemplary embodiment.

[0036] Figure 4 This is a flowchart illustrating another adaptive PID control method for a blood pressure simulator according to an exemplary embodiment. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0038] Example 1

[0039] This embodiment provides an implementation flow of an adaptive PID control method for a blood pressure simulator. This method aims to achieve high-precision, highly adaptive closed-loop control of the blood pressure simulator's pneumatic system by online identification of the controlled object model and intelligent tuning of control parameters.

[0040] Reference Figure 1This document illustrates an overall block diagram of an adaptive PID control device for a blood pressure simulator provided in an embodiment of this application. In one embodiment, the core of the entire adaptive PID control device is a main control module 10, which includes a PID controller, replacing the traditional fixed-parameter controller. The PID controller receives a pressure setpoint Ps and a pressure feedback value Pf from the controlled object, using the error e(t) between them as the main input, and outputs a control quantity u(t) to the actuator drive module 30. The actuator drive module 30 converts u(t) into a specific drive signal (such as a motor pulse or valve opening voltage), which acts on the controlled object actuator 40. The controlled object is the pneumatic system 50 of the blood pressure simulator, including an air pump, valves, air lines, and connected cuffs, etc. The adaptive PID control device integrates two key functional units: an online identification module 60 and an intelligent parameter tuning module 70. The online identification module 60 is responsible for sensing the dynamic characteristics of the controlled object in real time and transmitting the identified model parameters to the intelligent parameter tuning module 70. The intelligent parameter tuning module 70 calculates and updates the core parameters (proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd) required by the PID algorithm in real time based on these model parameters and the current error e(t) and error change rate ec(t), thereby giving the entire control loop adaptive capability.

[0041] Combined with reference Figure 2 The diagram illustrates in detail the specific steps of the method in this embodiment in the form of a flowchart.

[0042] Step S101: Online identification step, online identification of the approximate model parameters of the air pressure system of the blood pressure simulator under the current working conditions.

[0043] Specifically, such as Figure 3 As shown, step S101 further includes the following steps:

[0044] Step S201: Initialization. When the blood pressure simulator is powered on, its internal main control module loads a set of preset, relatively conservative default PID parameters. Simultaneously, the system loads the default operating mode, such as the blood pressure monitor's regular detection mode, and prepares to receive pressure setting commands from the host computer or user interface.

[0045] Step S202: Real-time data acquisition and error calculation. During operation, the pressure sensing module 20 continuously monitors the actual pressure within the pneumatic system and provides this pressure feedback value Pf to the main control module 10. The main control module 10 compares this pressure feedback value Pf with the user-set target pressure value Ps, calculating the current pressure error e(t) = Ps - Pf. Simultaneously, by performing a differential operation on the error e(t), the rate of change of the error ec(t) = de(t) / dt can be obtained. It can be understood that the two key variables, error e(t) and the rate of change of the error ec(t), constitute the basis for subsequent intelligent parameter tuning and control calculations.

[0046] Step S203: Online identification of approximate model parameters. This step aims to obtain the mathematical model of the pneumatic system 50 under the current specific operating conditions. It should be noted that this model will change with factors such as pressure level, cuff load, ambient temperature, and even component aging. To avoid interfering with normal pressure control, online identification is usually performed when the system reaches a certain pressure setpoint and enters a steady state or during periodic control gaps. Specifically, the main control module 10 superimposes a preset excitation signal on the control quantity u(t) required to maintain the current steady-state pressure. To minimize the impact of the identification process on the normal operation of the system, the amplitude of this excitation signal is usually small, for example, only about 1% of the total control quantity. As an optional implementation, the excitation signal can be a pseudo-random binary sequence, which has good autocorrelation characteristics and wide bandwidth characteristics, which is conducive to fully exciting the dynamic characteristics of the system. While applying the excitation signal, the pressure sensing module 20 records the small fluctuation sequence of the pressure feedback value Pf caused by the excitation at a high sampling rate. Subsequently, the online identification module 60 starts and runs the preset identification algorithm. In this embodiment, recursive least squares is used as the identification algorithm. It has a moderate computational load, is easy to implement on a microcontroller, and can update model parameters online in real time. Based on the input excitation signal sequence and the output pressure fluctuation sequence, recursive least squares can identify an approximate model of the pressure system 50 under the current operating conditions.

[0047] As a preferred embodiment, a first-order inertial plus pure time-delay model is chosen to approximate the dynamic characteristics of the pressure system 50, and its transfer function can be expressed as:

[0048] ;

[0049] Where K represents the system gain, reflecting the degree of influence of the control quantity change on the steady-state pressure value; T represents the system time constant, reflecting the speed of the system response; and τ represents the pure time delay, reflecting the delay from the issuance of the control quantity to the start of the system response. The final output of the online identification step S203 is the specific value of these three key model parameters (K, T, τ) identified under the current operating condition.

[0050] Step S102: Intelligent parameter tuning step, based on the approximate model and parameters obtained in the online identification step, uses an intelligent inference mechanism to calculate and update the parameters of the PID controller in real time.

[0051] Specifically, this step is by Figure 1 The intelligent parameter tuning module 70 shown completes its operation. In this embodiment, the intelligent parameter tuning module 70 is a fuzzy inference engine. The fuzzy inference engine receives the approximate model parameters (K, T, τ) from step S203 and the real-time error e(t) and error change rate ec(t) from step S202 as inputs. The intelligent parameter tuning module 70 internally stores a set of expert rule bases. These rules are established based on the knowledge and experience of control engineering experts and describe how to adjust the PID parameters to achieve optimal control performance under different system states. Based on the input information, the fuzzy inference engine calculates the required adjustment amounts ΔKp, ΔKi, and ΔKd for the three PID parameters in real time through a series of operations such as fuzzification, rule inference, and defuzzification. Then, the main control module 10 uses these adjustment amounts to update the current PID parameters: Kp_new = Kp_old + ΔKp, Ki_new = Ki_old + ΔKi, Kd_new = Kd_old + ΔKd.

[0052] Step S103: Adaptive control step, using the PID controller parameters updated in the intelligent parameter tuning step to perform closed-loop control of the air pressure system.

[0053] Specifically, the PID controller within the main control module 10 uses the PID parameters (Kp_new, Ki_new, Kd_new) updated in step S102, combined with the current error e(t) and error change rate ec(t), to calculate a new control quantity u(t) using the standard PID calculation formula. This control quantity u(t) is converted into a drive signal for the actuator 40 (e.g., a pulse width modulation signal controlling the speed of the air pump or a voltage signal controlling the opening of the exhaust proportional valve) through the actuator drive module 30, thereby precisely regulating the pneumatic system 50 and forming a complete adaptive closed-loop control.

[0054] To further clarify the working process of this embodiment, a stepped venting task is used as an example. This task requires the pressure to start at 300 mmHg and vent downwards in steps of 30 mmHg until a lower pressure value is reached. When the system stabilizes at the first pressure plateau of 300 mmHg, the system automatically performs online identification (step S101) and parameter tuning (step S102) to obtain the optimal PID parameters for this high-pressure range. When the venting command is issued and the target pressure becomes 270 mmHg, the system uses these optimized parameters for control, achieving a rapid pressure drop with almost no overshoot. After the pressure stabilizes at the 270 mmHg plateau, the system performs online identification and parameter tuning again to prepare the optimal PID parameters for the next pressure step (240 mmHg). This cycle repeats continuously, with the PID controller in its optimal "tailor-made" state for each pressure plateau, thus ensuring a smooth, rapid, and highly accurate step-by-step venting process over a wide pressure range (e.g., from 5 mmHg to 300 mmHg).

[0055] Example 2

[0056] This embodiment focuses on the specific implementation of the intelligent parameter tuning step S102 mentioned in Embodiment 1, particularly the expert rule base and intelligent inference mechanism within the intelligent parameter tuning module. In this embodiment, the intelligent inference mechanism is a fuzzy inference mechanism. As an effective intelligent control technology, fuzzy inference has the advantage of simulating the thinking of human experts and handling nonlinear and uncertain problems, making it very suitable for online tuning of PID parameters.

[0057] In this embodiment, the intelligent parameter tuning module (fuzzy inference engine) includes a fuzzification interface, a knowledge base (containing a database and a rule base), an inference engine, and a defuzzification interface. Its input variables are the normalized pressure error E and the error change rate EC. Normalization is the process of mapping the actual physical quantities e(t) and ec(t) to a fuzzy logic domain (usually [-1, 1] or a standardized range such as [-6, 6]) to facilitate processing using a unified membership function. The output variables are the adjustment amounts ΔKp, ΔKi, and ΔKd of the PID parameters.

[0058] The fuzzy subsets (linguistic variable values) of input variables E and EC are typically divided into seven levels: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large. Each fuzzy subset corresponds to a membership function, such as a triangular or Gaussian membership function, which describes the degree to which an input value belongs to that fuzzy subset.

[0059] The expert rule base is the core of fuzzy inference, consisting of a series of fuzzy conditional statements in the form of "IF-THEN". These rules embody the strategies employed by control experts to adjust PID parameters under different conditions. Below are example rules from the expert rule base in this embodiment, along with their detailed explanations:

[0060] Rule 1 (for rapid response to large errors): If the error E is large and the rate of change of error EC is zero, then the proportional coefficient adjustment ΔKp is large, the integral coefficient adjustment ΔKi is zero, and the derivative coefficient adjustment ΔKd is small. This rule is based on the following: when the actual system pressure Pf is much lower than the set pressure Ps (i.e., the error E is large), and the pressure change trend is not obvious (i.e., the rate of change of error EC is zero), it indicates that the system response is slow or in the initial startup phase. At this time, the proportional coefficient Kp should be significantly increased (ΔKp is large) to obtain a strong control action, allowing the system to quickly approach the set value. Simultaneously, to prevent excessive integral accumulation leading to severe overshoot (i.e., integral saturation) when the error is large, the integral action should not be increased temporarily (ΔKi is zero). Appropriately increasing the derivative action (ΔKd is small) helps predict future trends and improves system stability.

[0061] Rule 2 (for suppressing overshoot): If the error E is small positive and the rate of change of error EC is large negative, then the proportional coefficient adjustment ΔKp is small negative, the integral coefficient adjustment ΔKi is small positive, and the derivative coefficient adjustment ΔKd is large positive. The logic of this rule is that when the actual system pressure Pf is very close to the set pressure Ps (i.e., the error E is small positive), but the pressure is rapidly rising and approaching the set value (i.e., the rate of change of error EC is large negative because e(t) is decreasing), the system is highly likely to overshoot. To effectively "brake," the proportional action should be reduced (ΔKp is small negative) to avoid excessive control; simultaneously, the derivative action needs to be significantly enhanced (ΔKd is large positive), because the derivative term is most sensitive to the rate of change of error, and a strong derivative action can produce a strong reverse control effect, effectively suppressing overshoot. At this point, the error is already very small, and the integral action (ΔKi is small positive) can be appropriately increased to prepare for subsequent elimination of steady-state error.

[0062] Rule 3 (for fine steady-state control): If the error E is zero and the rate of change of error EC is zero, then the proportional gain adjustment ΔKp is zero, the integral gain adjustment ΔKi is positively small, and the derivative gain adjustment ΔKd is zero. The logic of this rule is that when the system pressure has stabilized near the setpoint (i.e., both the error E and the rate of change of error EC are close to zero), it indicates that the system has entered a steady state or quasi-steady state. At this time, the proportional and derivative actions should remain unchanged (ΔKp and ΔKd are zero) to maintain system stability. However, since the system may have a steady-state error (i.e., a small, persistent error), it is necessary to gradually eliminate this steady-state error by slightly increasing the integral action (ΔKi is positively small), thereby improving the final control accuracy.

[0063] Rule 4 (Mode-Specific Rule): If the current operating mode is stepped venting mode, increase the overall weight of the differential coefficient Kd and decrease the overall weight of the integral coefficient Ki. It should be noted that this rule is not a direct IF-THEN control rule, but rather a meta-adjustment strategy for the entire rule base or defuzzification process, used to adapt to different operating modes. The stepped venting mode is characterized by a sudden change in the pressure target value. Under this condition, the system's primary task is to respond quickly to the step signal and suppress overshoot. Therefore, when entering this mode, the fuzzy inference engine automatically adjusts its internal parameters, for example, by increasing the magnitude of ΔKd for all rule outputs and decreasing the magnitude of ΔKi through a weighting factor. Enhancing the differential action (Kd) improves the system's response speed and predictability to sudden changes, while weakening the integral action (Ki) effectively avoids integral saturation caused by large errors in the initial stage of a pressure step, thus preventing significant overshoot. This adjustment reflects the dynamic adjustment of the control target performance under different operating modes.

[0064] The following example of a large-scale pressurization task illustrates its working process. Assume the system needs to perform a pressurization from 50 mmHg to 180 mmHg. Initially, the pressure is 50 mmHg, the target is 180 mmHg, and the error e(t) is large, resulting in a positive value after normalization. At this point, the system primarily triggers rule 1, Kp is rapidly increased, the air pump operates at maximum power, and the pressure rises quickly. When the pressure approaches 180 mmHg, for example, reaching 175 mmHg, the error e(t) decreases, and the normalized value of E may be small positive. Simultaneously, because the pressure is still rising rapidly, the error is decreasing rapidly, and ec(t) becomes negative, potentially resulting in a large negative value after normalization. At this point, the system primarily triggers rule 2, Kp begins to decrease, while Kd significantly increases, and the control action begins to "decelerate" to prevent the pressure from exceeding 180 mmHg. When the pressure fluctuates slightly around 180 mmHg and eventually stabilizes, both E and EC approach zero. The system triggers rule 3, which eliminates any possible static error of ±0.1 mmHg by fine-tuning Ki, ultimately stabilizing the pressure precisely at 180 mmHg.

[0065] Example 3

[0066] This embodiment focuses on another technical feature proposed in this application, namely, the performance evaluation and self-learning mechanism. This mechanism aims to endow the blood pressure simulator with long-term self-optimization capabilities to compensate for performance degradation caused by factors such as aging of equipment components and slow changes in the environment.

[0067] For example, such as Figure 4 As shown, step S104: Performance evaluation and self-learning step, periodically evaluates the performance indicators of the control process, and when the performance indicators exceed the preset performance range, triggers the self-learning process to fine-tune the preset control strategy on which the intelligent inference mechanism is based. The preset control strategy on which the intelligent inference mechanism is based is an expert rule base, and the self-learning process includes fine-tuning this expert rule base.

[0068] Step S104 is by Figure 1 The performance evaluation and self-learning module 80 shown is now complete. This module includes a performance evaluation submodule and a learning algorithm submodule. The performance evaluation submodule is used to quantitatively evaluate control performance. For example, after completing a full stepped venting process, the performance evaluation submodule calculates key performance indicators such as the maximum overshoot and average settling time throughout the process. Then, it compares these calculated current performance indicators with a set of preset performance thresholds or benchmarks in the system. These thresholds represent the performance level that the equipment should achieve under ideal conditions or at the time of manufacture.

[0069] When the comparison results of the performance evaluation submodule show that one or more current performance indicators are worse than the preset threshold (i.e., exceed the preset performance range), the self-learning process will be triggered, activating the learning algorithm submodule.

[0070] The learning algorithm submodule executes the corresponding learning algorithm to correct the control strategy based on the evaluation results. It executes the appropriate learning algorithm to correct the root cause of the performance degradation based on the specific information about "performance degradation" provided by the performance evaluation submodule (e.g., whether the settling time has increased or the overshoot has increased). In one specific embodiment of this application, the learning algorithm can be a simple rule weight adjustment algorithm, or a more complex gradient descent method or neural network algorithm. The learning algorithm submodule corrects the expert rule base or approximate model parameters, fine-tuning and updating both. This update is small and gradual to ensure system stability. The optimized strategy, i.e., the updated expert rule base or approximate model parameters, is fed back to the main control module, directly affecting the behavior of its internal adaptive PID controller, thereby improving performance in subsequent control tasks.

[0071] The working principle of this self-learning mechanism is illustrated below through a specific work scenario: Suppose a blood pressure simulator using the technology described in this application, at the time of manufacture, has an average settling time of 2.5 seconds for performing a standard stepped deflation task (e.g., from 200 mmHg to 100 mmHg, in 20 mmHg increments), and the system's preset performance threshold is 2.8 seconds. After one year of continuous use at the customer's site, the proportional solenoid valve used for deflation undergoes some changes in response characteristics due to mechanical wear, resulting in a slight delay in its opening and closing actions. This slow aging process causes the system's dynamic performance to gradually decline. At this point, when the device performs the same standard stepped deflation task again, the performance evaluation submodule calculates after the task that the current average settling time has increased to 3.0 seconds. The performance evaluation submodule compares 3.0 seconds with the preset threshold of 2.8 seconds and finds that the current performance exceeds the acceptable range, thus triggering the self-learning process. The learning algorithm submodule is activated. It analyzed the historical performance database (a storage area for historical performance indicators) and found that the performance degradation was mainly reflected in the increased settling time, which was particularly noticeable when the pressure stepped from high to low. This indicated that the problem might lie in the exhaust response. Based on this analysis, the learning algorithm submodule decided to fine-tune the rules in the expert rule base related to "large negative error" (corresponding to the exhaust process). Specifically, it might shift the center point of the membership function of the output ΔKp towards "large positive" by 5% when processing rules like "if E is large negative...". The physical meaning of this adjustment is that when the system detects the need for large exhaust, it increases the proportional action earlier and more strongly to compensate for the impact of the proportional valve response delay. After the fine-tuning, the updated expert rule base was saved and immediately applied to the next control task. When the device performed the same standard stepped exhaust task again, because the control strategy had been self-optimized, its average settling time recovered to about 2.6 seconds, returning to an acceptable performance range.

[0072] Example 4

[0073] This embodiment provides a blood pressure simulator for implementing the above-described adaptive PID control method, the specific hardware architecture of which is described below. Please refer to... Figure 1 This is an overall block diagram of an adaptive PID control device for a blood pressure simulator provided in an embodiment of this application.

[0074] The core of this device is the main control module 10. In this embodiment, the main control module 10 can employ a high-performance 32-bit microcontroller, such as one based on the ARM Cortex-M4 core. The advantage of using such a microcontroller is that it typically has a built-in hardware floating-point unit, which can efficiently execute the recursive least squares method, fuzzy logic operations, and the large number of floating-point calculations involved in the PID control algorithm, ensuring the real-time performance of the entire control system. The functional modules described in embodiments 1 to 3, such as the online identification module 60, the intelligent parameter tuning module 70 (fuzzy inference engine), the performance evaluation submodule, and the learning algorithm submodule, are all embedded as software programs (firmware) in the internal flash memory of the main control module 10. The main control module 10 also includes a PID controller, whose parameters (Kp, Ki, Kd) are updated in real-time by the intelligent parameter tuning module 70 (fuzzy inference engine).

[0075] The pressure sensing module 20 is used to detect the pressure within the pneumatic system in real time. In this embodiment, a piezoresistive pressure sensor with good linearity and temperature compensation characteristics can be selected, whose range covers the typical operating range of the blood pressure simulator (e.g., 0-400 mmHg). The sensor output is typically a weak analog voltage signal. The pressure sensing module 10 first amplifies and filters the sensor signal to improve the signal-to-noise ratio, and then digitizes it through a high-precision analog-to-digital converter. To ensure the accuracy of the pressure feedback value and thus achieve high-precision control, a 24-bit high-precision Σ-Δ analog-to-digital converter is preferably used here. The digitized pressure data is transmitted to the main control module 10 as the pressure feedback value Pf.

[0076] The actuator 40 is the component that actually regulates the air pressure. In this embodiment, the air pressure system 50 consists of two parts: pressurization and exhaust. The pressurization part can use a precision lead screw piston pump driven by a stepper motor. By precisely controlling the number of rotation steps of the stepper motor, the cylinder volume is changed, thereby achieving precise pressurization of the air pressure. The exhaust part can use a high-frequency response proportional solenoid valve, whose opening degree is proportional to the input control voltage, enabling continuous adjustment of the exhaust rate.

[0077] The actuator drive module 30 connects the main control module 10 and the actuator 40, and is responsible for converting control signals into drive signals. It receives logic-level control signals from the main control module 10 and converts them into high-current or high-voltage signals sufficient to drive the actuator 40. For example, it may include a stepper motor driver chip that receives pulse and direction signals from the main control module 10 to control the stepper motor; it may also include a digital-to-analog converter and a power amplifier circuit to convert the digital control values ​​calculated by the main control module 10 into the analog voltage required to drive the proportional solenoid valve.

[0078] The communication interface 90 is used for data exchange between the device and external equipment (such as a host computer used for calibration and setup). It can be a general-purpose USB interface or an industrially common RS-485 interface. Through this interface, users can set the target pressure Ps, select the operating mode, read real-time pressure data, and view the device status.

[0079] The overall workflow of the device is described as follows: The main control module 10, acting as the brain of the system, continuously runs the main program loop. It reads the current pressure feedback value Pf through the pressure sensing module 20 and compares it with the pressure setpoint Ps received through the communication interface 90. Based on the error e(t) and the error change rate ec(t), the software modules inside the main control module 10 work together: the online identification module 60 identifies the approximate model parameters of the system at appropriate times, the intelligent parameter tuning module 70 tunes the PID parameters in real time according to the approximate model parameters and the error state, and the updated PID controller calculates the final control quantity u(t). This control quantity is sent to the actuator drive module 30, which converts it into precise drive signals for the stepper motor and proportional valve, thereby adjusting the action of the actuator and changing the air pressure. After detecting the pressure change, the pressure sensing module 20 transmits the new pressure feedback value Pf back to the main control module 10, thus forming a complete, high-speed, and intelligent adaptive closed-loop control system.

[0080] The terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0081] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing application concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions claimed in this application.

Claims

1. An adaptive PID control method for a blood pressure simulator, characterized in that, include: Online identification steps: Identify the approximate model parameters of the pneumatic system of the blood pressure simulator under the current operating conditions online; Intelligent parameter tuning steps: Based on the approximate model parameters obtained in the online identification step, the PID controller parameters are calculated and updated in real time using an intelligent inference mechanism; and Adaptive control step: Use the updated PID controller parameters from the intelligent parameter tuning step to perform closed-loop control of the air pressure system.

2. The adaptive PID control method for a blood pressure simulator as described in claim 1, characterized in that, Also includes: Performance evaluation and self-learning steps: Periodically evaluate the performance indicators of the control process, and when the performance indicators exceed the preset performance range, trigger the self-learning process to fine-tune the preset control strategy on which the intelligent reasoning mechanism is based.

3. The adaptive PID control method for a blood pressure simulator as described in claim 2, characterized in that, The intelligent reasoning mechanism is based on a preset control strategy of an expert rule base, and the self-learning process includes fine-tuning the expert rule base.

4. The adaptive PID control method for a blood pressure simulator as described in claim 1, characterized in that, The online identification steps specifically include: A preset excitation signal is superimposed on the control quantity of the actuator drive module; and Based on the excitation signal and pressure output data, the approximate model parameters are identified using the recursive least squares method.

5. The adaptive PID control method for a blood pressure simulator as described in claim 4, characterized in that, The approximate model is a first-order inertial plus pure time-delay model: ； Where K represents the system gain, reflecting the degree of influence of the control quantity change on the steady-state pressure value; T represents the system time constant, reflecting the speed of the system response; and τ represents the pure time delay, reflecting the delay from the issuance of the control quantity to the start of the system response.

6. The adaptive PID control method for a blood pressure simulator as described in claim 1, characterized in that, The intelligent reasoning mechanism is a fuzzy reasoning mechanism.

7. An adaptive PID control device for a blood pressure simulator, comprising a main control module, a pressure sensing module, and an actuator drive module, characterized in that, Also includes: The online identification module is configured to identify the approximate model parameters of the air pressure system of the blood pressure simulator online; The intelligent parameter tuning module is configured to calculate and update the PID controller parameters in real time based on the approximate model parameters using an intelligent inference mechanism. The main control module includes a PID controller, which is configured to set the PID controller using the updated PID controller parameters from the intelligent parameter tuning module, and then perform closed-loop control of the pneumatic system through the PID controller and the actuator drive module.

8. The apparatus according to claim 7, characterized in that, Also includes: The performance evaluation and self-learning module is configured to periodically evaluate the performance indicators of the control process, and when the performance indicators exceed the preset range, trigger the self-learning process to fine-tune the preset control strategy on which the intelligent parameter tuning module is based.

9. The apparatus according to claim 7, characterized in that, The intelligent parameter tuning module is a fuzzy inference engine, including a fuzzification interface, a knowledge base (containing a database and a rule base), an inference engine, and a defuzzification interface. Its inputs are the normalized error and the error change rate, and its output is the adjustment amount of the PID parameters.

Images