Drug supply control system and method for anesthetic machine in anesthesiology department

Abstract

The present invention relates to the technical field of drug dosage supply control, and discloses a drug dosage supply control system and method for an anesthetic machine in the anesthesiology department, including: smoothing sensor data through weighted moving average filtering to reduce measurement noise; subsequently, calculating the patient's individualized basal infusion rate based on a pharmacokinetic model to provide reasonable feedforward control; combining PID feedback control with feedforward compensation, adjusting the control signal in real time through errors to enhance dynamic adaptability; and further optimizing the control signal by using optimal state feedback compensation based on LQR to improve the system robustness, calculating the optimal compensation amount, and superimposing it on the PID control signal; introducing fault detection and fault tolerance compensation throughout the process to identify abnormal situations, adjust the control strategy, ensure system stability, and achieve precise control of anesthetic drugs.

Description

Technical Field

[0001] The present invention relates to the technical field of drug supply control, and specifically to a drug supply control system and method for an anesthetic machine used in the anesthesiology department. Background Art

[0002] Currently, in the field of anesthetic drug supply control, traditional methods mainly rely on simple open-loop infusion or a single feedback controller to adjust the drug infusion rate. However, due to large fluctuations in physiological parameters, significant interference from sensor signal noise, and high patient individual differences, these methods are difficult to achieve precise, stable, and continuous drug supply.

[0003] The common PID feedback control in the prior art can achieve a certain degree of error compensation, but it is sensitive to noise, has problems of integral saturation and insufficient handling of system nonlinear responses, resulting in difficult elimination of steady-state errors and overshoot and oscillation problems in dynamic responses. In addition, although some advanced technologies based on model predictive control (MPC) can predict the system state in advance and optimize the control input to a certain extent, they are difficult to promote in practical applications due to inaccurate model parameter estimation, complex online optimization calculations, and insufficient real-time performance.

[0004] Therefore, this solution proposes a multi-level closed-loop control method that combines feedforward, PID feedback, deterministic state feedback compensation, and fault detection and fault-tolerant compensation. This method has obvious advantages over the existing single feedback or MPC control technologies in terms of stability, real-time performance, and adaptability; it realizes continuous closed-loop control from data preprocessing to feedforward reference determination, and then to feedback control and optimal compensation, significantly improving the accuracy and safety of drug supply. Summary of the Invention

[0005] The present invention provides a drug supply control system and method for an anesthetic machine used in the anesthesiology department, which helps to solve the problems mentioned in the above background art.

[0006] The present invention provides the following technical solution: A drug supply control system for an anesthetic machine used in the anesthesiology department, comprising:

[0007] A multi-dimensional sensor module: for collecting the anesthetic drug concentration in the patient's body.

[0008] A data acquisition and clock synchronization module: for reading sensor data in real time and recording the sampling time simultaneously to ensure the accuracy of the data timestamp.

[0009] A data preprocessing module: using weighted moving average filtering technology to smooth the collected raw data and output the smoothed data as the input for subsequent control.

[0010] Patient Physiological Parameter Database: Stores basic patient information and physiological indicators, providing individualized parameters for obtaining the feedforward control output;

[0011] Fault Detection and Fault Tolerance Module: Monitors the data of each module in real time, detects anomalies through statistical methods, and determines whether to activate the fault tolerance mechanism;

[0012] Drug Dosage Control Module: Used to receive the drug dosage control signal and adjust the injection rate of the anesthetic drug input into the patient's body according to the control signal.

[0013] A control method for a drug supply control system using an anesthetic machine in the anesthesiology department, including:

[0014] During the anesthesia of the patient, collect the anesthetic drug concentration in the patient's body;

[0015] Filter the collected anesthetic drug concentration of the patient;

[0016] According to the target concentration of the anesthetic drug in the patient's body, obtain the feedforward control output that makes the anesthetic drug concentration in the patient's body reach the target concentration;

[0017] Perform feedback correction based on the target concentration and the filtered anesthetic drug concentration;

[0018] Perform optimization compensation according to the error state to obtain the compensated infusion rate of the anesthetic drug input into the patient's body;

[0019] During the process of inputting anesthetic drugs into the patient's body, evaluate the anomaly score of the data collected by the sensor;

[0020] When the result of the anomaly score evaluation shows an anomaly, perform intelligent compensation.

[0021] Optionally, the filtering process of the collected anesthetic drug concentration of the patient includes:

[0022] The filtering process of the anesthetic drug concentration in the patient's body collected is specifically:

[0023] Record the anesthetic drug concentration of the patient collected at time as ;

[0024] Set sliding window;

[0025] According to the set sliding window, collect the anesthetic drug concentrations in the most recent 5 unit times and record them as , , , , ;

[0026] Set weights for the anesthetic drug concentration at each moment:

[0027] , , , ,

[0028] where is the weight at the current moment , is the weight at the previous moment , and so on; for the setting of the physiological parameter weights, it can be adjusted as needed. To make the data at the latest moment more representative, the setting of the weights needs to satisfy ≥ ≥ ≥ ≥ >0;

[0029] Calculate the weighted moving average of the anesthetic drug concentration collected in the patient's body:

[0030]

[0031] where:

[0032] : represents the anesthetic drug concentration in the patient's body obtained after filtering at time ;

[0033] : represents the anesthetic drug concentration of the patient collected at time , ;

[0034] : represents the preset weight of the anesthetic drug concentration at the corresponding moment.

[0035] Optionally, obtaining the feedforward control output that enables the anesthetic drug concentration in the patient's body to reach the target concentration according to the target concentration of the anesthetic drug in the patient's body includes:

[0036] Set the target concentration of the anesthetic drug in the patient's body, denoted as ;

[0037] Then when the anesthetic drug concentration in the patient's body is required to be , the infusion rate of the anesthetic drug required to be input into the patient's body is:

[0038]

[0039] where:

[0040] : Distribution volume, determined according to the patient's weight, with a value range of ;

[0041] : Drug elimination rate constant, unit , with a value range of ;

[0042] : Feedforward control output, representing the infusion rate of anesthetic drugs, unit: .

[0043] Optionally, perform the feedback correction:

[0044] Calculate the anesthetic drug concentration error:

[0045]

[0046] Where:

[0047] : Is the anesthetic drug concentration in the patient's body obtained after filtering;

[0048] : Is the target concentration of anesthetic drugs set for the patient's body;

[0049] Calculate the feedback correction amount:

[0050]

[0051] Where:

[0052] : Is the feedback correction, representing the infusion rate of anesthetic drugs, unit: ;

[0053] : Represents the drug concentration error;

[0054] : Are the proportional, integral, and differential gains respectively, with a value range of , , ; And obtain the integral term and the differential term ;

[0055] Combine the feedforward control input and the feedback correction:

[0056]

[0057] Where, is the infusion rate of anesthetic drugs input into the patient's body after correction.

[0058] Optionally, the optimizing compensation according to the error state to obtain the infusion rate of the anesthetic drug input into the patient's body includes:

[0059] Establishing a drug dynamic model:

[0060]

[0061] Wherein: is the target concentration of the anesthetic drug in the patient's body, equal to ; is the feedforward control output, equal to ;

[0062] Defining the error:

[0063]

[0064] Linearizing the disturbance to obtain the error dynamics:

[0065]

[0066] At , introducing a deviation variable:

[0067]

[0068] Substituting gives:

[0069]

[0070] At steady state, , that is , and finally obtaining:

[0071]

[0072] Establishing a state equation, letting the state :

[0073]

[0074] Then we get , ;

[0075] Setting the state weight and the control weight , and constructing the cost function as:

[0076]

[0077] is the state weight, set artificially, and the larger the value, the more strict the punishment for the error;

[0078] To control the weight, it is set manually. The larger the value, the smaller the change in the input that is desired to be controlled;

[0079] Through the constructed cost function, the algebraic Riccati equation obtained is:

[0080]

[0081] Solve the algebraic Riccati equation to obtain a positive definite solution;

[0082] Calculate the optimal feedback gain :

[0083]

[0084] Then the optimal compensation control is:

[0085]

[0086] Finally, the optimal control at the current moment is obtained as:

[0087]

[0088] Among them, is the infusion rate of the anesthetic drug finally determined to be input into the patient's body.

[0089] Optionally, during the process of inputting the anesthetic drug into the patient, an abnormal score evaluation is performed on the data collected by the sensor, including:

[0090] During the control process, an abnormal score evaluation is performed:

[0091]

[0092] Among them:

[0093] : The abnormal score at time ;

[0094] : The number of sensors for obtaining the patient's physiological parameters;

[0095] : The th sensor's measurement value at time ;

[0096] : The th sensor's historical mean value;

[0097] : The The standard deviation of each sensor;

[0098] : The weights of each sensor, with a value range of ;

[0099] Set the abnormal determination threshold to ;

[0100] At each moment, the obtained is compared with ;

[0101] If at a certain moment , it is determined that an abnormality exists.

[0102] Optionally, when the result of the abnormal score evaluation shows an abnormality, intelligent compensation is performed, including:

[0103] When an abnormality exists, intelligent compensation is performed, and the compensation amount is , and the calculation formula is:

[0104]

[0105] Where:

[0106] : Compensation adjustment amount;

[0107] : Sensitivity adjustment coefficient, with a value range of ;

[0108] : Hyperbolic tangent function, used to smooth the compensation output and prevent excessive compensation.

[0109] The present invention has the following beneficial effects:

[0110] 1. By organically combining the three algorithms of feedforward control, PID feedback, and LQR compensation, its greatest beneficial effect lies in achieving a multi-level and multi-angle control closed-loop. Each layer of the algorithm makes full use of the output data of the previous layer as input, and gradually reduces the system error through optimization and feedback compensation, thereby achieving precise control of the anesthetic drug supply. Specifically, the feedforward control calculates the steady-state reference based on the pharmacokinetic model, providing a deterministic reference for the system, so that the subsequent feedback regulation does not start from scratch, but makes fine-tuning around a reasonable reference. Then, the PID feedback control with feedforward compensation calculates the error using real-time measurement data and quickly compensates for dynamic disturbances through proportional, integral, and differential regulation, ensuring that the system can respond quickly in a short time while eliminating long-term deviations. Finally, the LQR compensation further utilizes the linearized model and optimal control theory to solve the optimal feedback gain by constructing a quadratic cost function, deeply compensating for the error dynamics, so that the entire system maintains a stable response in the future time domain. The advantage of this algorithmic involvement is that each part works together, making the control signal have both the predictability of feedforward and the real-time correction ability of feedback. At the same time, the optimal compensation part ensures that a near-optimal control effect can still be obtained when the system is disturbed or the model is inaccurate. Compared with traditional methods such as using PID control or MPC alone, this multi-level control scheme can better handle dynamic changes and noise interference in complex environments, greatly improving the robustness, response speed, and stability of the system, thereby significantly reducing the risks during clinical anesthesia and ensuring patient safety and treatment effects.

[0111] 2. The weighted moving average filtering method is used to process the raw data collected by the sensor. Its main advantage is that it can effectively smooth the random noise in the short term and retain the main trend of the data, thus providing a more stable and reliable input for subsequent control. By presetting weights, this method assigns higher weights to the latest data and lower weights to the older data, which can not only fully reflect the current changes but also suppress the instantaneous fluctuations caused by measurement noise. Using historical data statistics to determine the weights can make the filtering result have strong certainty and avoid the fluctuation problems caused by the unstable update of the uncertainty matrix in the dynamic recursive algorithm. Compared with the traditional simple average method, this filtering technology has significant improvements in real-time performance and accuracy, and at the same time has a small amount of calculation, which is convenient for embedding in the real-time control system. In addition, the weighted moving average method is simple to implement, easy to realize, and has low hardware requirements, showing good adaptability and robustness in the preprocessing of multi-channel sensor data, thus providing a solid foundation for subsequent feedforward control, PID feedback regulation, and optimal state feedback compensation based on pharmacokinetics. This method can solve the problem of unstable control signals caused by sensor noise in the existing technology, making the subsequent controller generate control signals based on more reliable data input, further ensuring the response speed and stability of the entire system, and thus significantly improving the accuracy of anesthetic drug supply and patient safety.

[0112] 3. By calculating the feedforward control of the influence of drug distribution and elimination, its core idea is to calculate the basic drug infusion rate required to maintain the target concentration under steady-state conditions based on the pharmacokinetic model. This method can pre-estimate the drug demand under ideal conditions, thus providing a definite benchmark at the start of the system. This method combines the target concentration, drug elimination rate, and volume of distribution organically through the formula to make the calculation result more in line with the actual situation. This feedforward control mechanism can provide compensation before the system is disturbed, reducing the possibility of error generation, thus making the entire control process have higher steady-state performance. At the same time, the calculation process of feedforward control is simple and clear, and the parameters are all obtained through offline experiments and clinical data, with high certainty and reliability. Compared with the traditional method that solely relies on feedback regulation, this feedforward control scheme can compensate for system defects in advance before the interference arrives, reducing the response time and control error required for feedback regulation, thus greatly improving the accuracy and safety of anesthetic drug supply, providing a good initial benchmark for subsequent PID feedback control, making the entire control closed-loop smoother, faster, and more predictive, and thus effectively solving the problems of large initial errors and system response lags caused by the lack of feedforward control in the existing technology.

[0113] 4. In the PID feedback control with feedforward compensation, by combining the benchmark obtained from feedforward control with the current error Calculated feedback correction amount Add them together to form an overall control signal ; This design not only utilizes the steady-state reference provided by feedforward control but also instantaneously adjusts the real-time error through PID feedback, thereby achieving a fast response to dynamic disturbances and eliminating steady-state errors; the proportional gain directly amplifies the current error, enabling the system to respond rapidly; the integral term ensures that long-term deviations are eliminated; the derivative term predicts the trend of error change, which helps to suppress system overshoot and oscillation; through this combination of feedforward and feedback, a basic compensation can be achieved before the disturbance occurs, and at the same time, it can be rapidly adjusted after the disturbance appears, making the response of the entire control system to errors smoother and more stable; this design not only overcomes the problems of integral saturation and overshoot that may occur in traditional PID control when responding to dynamic disturbances, but also, since the feedback regulation is calculated based on real-time smooth data, its robustness and anti-interference ability have been significantly improved, thus making the control of anesthetic drug supply more accurate, real-time, and with high safety, solving the problem of concentration fluctuations caused by lagged response in existing single feedback control methods and greatly reducing the patient risk.

[0114] 5. Use the LQR method to perform optimal state feedback compensation on the system error. Its main advantage lies in comprehensively considering the state error and the cost of control input by constructing a quadratic cost function, thereby obtaining an optimal feedback gain ; This method first linearizes and approximates the system pharmacokinetic model and writes the error dynamics as a state-space model , where represents the error between the target concentration and the actual concentration, and this model fully reflects the dynamic characteristics of the system under small disturbance conditions; after constructing the cost function , a positive definite matrix is obtained by solving the algebraic Riccati equation, and then the optimal feedback gain is calculated; using this feedback gain, the control compensation law It can quickly zero the error while ensuring a smooth change in the control input. The beneficial effect of this method is that it not only provides a theoretically optimal compensation control increment, but also, since all parameters are determined by the physical model, it has a high degree of certainty and robustness. LQR compensation can effectively solve the dynamic response problem that is difficult to eliminate in traditional PID feedback control, significantly reducing the risk of system overshoot and oscillation, making the final control signal more accurate and stable. In addition, this method realizes the automatic adjustment of the control force through quadratic cost function optimization, which helps to maintain the system performance under various interference conditions, thus greatly improving the safety and control accuracy during the anesthesia drug supply process and effectively making up for the limitations existing when using PID feedback control or feedforward control alone in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0115] Figure 1 It is a schematic diagram of the basic process of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0116] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0117] Embodiment 1, A drug supply control system for an anesthesia machine in the anesthesiology department, comprising:

[0118] Multidimensional sensor module: used to collect the concentration of anesthetic drugs in the patient's body;

[0119] Data acquisition and clock synchronization module: used to read sensor data in real time and record the sampling time simultaneously to ensure the accuracy of the data timestamp;

[0120] Data preprocessing module: uses weighted moving average filtering technology to smooth the collected raw data and outputs the smoothed data as the input for subsequent control;

[0121] Patient physiological parameter database: stores the patient's basic information and physiological indicators to provide individualized parameters for obtaining the feedforward control output;

[0122] Fault detection and fault tolerance module: monitors the data of each module in real time, detects abnormalities through statistical methods, and determines whether to start the fault tolerance mechanism;

[0123] Drug dosage control module: used to receive the drug dosage control signal and adjust the injection rate of the anesthetic drug input into the patient's body according to the control signal.

[0124] Embodiment 2, Refer to Figure 1, a control method for a drug supply control system using an anesthetic machine in the anesthesiology department, including:

[0125] During the anesthesia of a patient, collect the anesthetic drug concentration in the patient's body;

[0126] Filter the collected anesthetic drug concentration of the patient;

[0127] According to the target concentration of the anesthetic drug in the patient's body, obtain the feedforward control output that enables the anesthetic drug concentration in the patient's body to reach the target concentration;

[0128] Perform feedback correction according to the target concentration and the filtered anesthetic drug concentration;

[0129] Perform optimization compensation according to the error state, and obtain the infusion rate of the anesthetic drug input into the patient's body after compensation;

[0130] During the process of inputting anesthetic drugs into the patient's body, perform abnormal score evaluation on the data collected by the sensor;

[0131] When the result of the abnormal score evaluation shows an abnormality, perform intelligent compensation. By organically combining the three algorithms of feedforward control, PID feedback, and LQR compensation, its greatest beneficial effect is to achieve a multi-level and multi-angle control closed-loop. Each layer of the algorithm makes full use of the output data of the previous layer as input, and gradually reduces the system error through optimization and feedback compensation, thereby achieving precise control of the anesthetic drug supply; specifically, the feedforward control calculates the steady-state benchmark based on the pharmacokinetic model, providing a deterministic reference for the system, so that the subsequent feedback regulation does not start from scratch, but fine-tunes around a reasonable benchmark; then, the PID feedback control with feedforward compensation calculates the error using real-time measurement data and quickly compensates for dynamic disturbances through proportional, integral, and derivative regulation, so as to ensure that the system can respond quickly in a short time and eliminate long-term deviations at the same time; finally, the LQR compensation further uses the linearized model and the optimal control theory to solve the optimal feedback gain by constructing a quadratic cost function, and deeply compensates for the error dynamics, so that the entire system maintains a stable response in the future time domain; the advantage of this algorithmic involvement is that each part works together, making the control signal have both the predictability of feedforward and the real-time correction ability of feedback, and at the same time the optimal compensation part ensures that the system can still obtain a near-optimal control effect even when the system is disturbed or the model is inaccurate; compared with using traditional methods such as PID control or MPC alone, this multi-level control scheme can better handle the dynamic changes and noise interference in a complex environment, greatly improving the robustness, response speed, and stability of the system, thereby significantly reducing the risks during clinical anesthesia and ensuring patient safety and treatment effects.

[0132] The filtering process for the collected anesthetic drug concentration of the patient includes:

[0133] Filter the anesthetic drug concentration in the patient's body collected, specifically:

[0134] Record the anesthetic drug concentration of the patient collected at time as ;

[0135] Set a sliding window of .

[0136] According to the set sliding window, collect the anesthetic drug concentrations in the recent 5 unit times and record them as , , , , ;

[0137] Set weights for the anesthetic drug concentration at each time respectively:

[0138] , , , ,

[0139] where is the weight at the current time , is the weight at the previous time , and so on; for the setting of the weights of physiological parameters, it can be adjusted as needed. To make the data at the latest time more representative, the setting of the weights needs to satisfy ≥ ≥ ≥ ≥ > 0;

[0140] Calculate the weighted moving average of the anesthetic drug concentration in the patient's body collected:

[0141]

[0142] where:

[0143] : represents the anesthetic drug concentration in the patient's body obtained after filtering at time ;

[0144] : represents the anesthetic drug concentration of the patient collected at time , ;

[0145] : The preset weight representing the concentration of anesthetic drugs at the corresponding moment. The weighted moving average filtering method is used to process the raw data collected by the sensor. Its main advantage lies in being able to effectively smooth the random noise in the short term and retain the main trend of the data, thus providing a more stable and reliable input for subsequent control; this method assigns a higher weight to the latest data and a lower weight to the older data through a preset weight, which can not only fully reflect the current changes but also suppress the instantaneous fluctuations caused by measurement noise; determining the weight using historical data statistics can make the filtering result have strong certainty and avoid the fluctuation problems caused by the unstable update of the uncertainty matrix in the dynamic recursive algorithm; this filtering technology has significant improvements in real-time performance and accuracy compared to the traditional simple average method, and at the same time has a small computational amount, making it easy to be embedded in the real-time control system; in addition, the weighted moving average method is simple to implement, easy to realize, and has low hardware requirements, showing good adaptability and robustness in the preprocessing of multi-channel sensor data, thus providing a solid foundation for subsequent feedforward control based on pharmacokinetics, PID feedback regulation, and optimal state feedback compensation. This method can solve the problem of unstable control signals caused by sensor noise in the existing technology, enabling the subsequent controller to be based on more reliable data input when calculating errors and generating control signals, further ensuring the response speed and stability of the entire system, and thus significantly improving the accuracy of anesthetic drug supply and patient safety.

[0146] Obtaining the feedforward control output that enables the concentration of anesthetic drugs in the patient's body to reach the target concentration according to the target concentration of anesthetic drugs in the patient's body includes:

[0147] Set the target concentration of anesthetic drugs in the patient's body, denoted as ;

[0148] Then when the concentration of anesthetic drugs in the patient's body is required to be , the infusion rate of anesthetic drugs required to be input into the patient's body is:

[0149]

[0150] Where:

[0151] : The volume of distribution, determined according to the patient's weight, with a value range of ;

[0152] : The drug elimination rate constant, unit , with a value range of ;

[0153] : The feedforward control output, representing the infusion rate of anesthetic drugs, unit: . By performing feedforward control calculations on the effects of drug distribution and clearance, the core idea is to calculate the basal drug infusion rate required to maintain the target concentration under steady-state conditions based on a pharmacokinetic model. This method can pre-estimate the drug requirements under ideal conditions, thus providing a definite benchmark at the start of the system; this method combines the target concentration, drug elimination rate, and volume of distribution through the formula , making the calculation results more in line with the actual situation; this feedforward control mechanism can provide compensation before system disturbances occur, reducing the possibility of error generation, thereby enabling the entire control process to have higher steady-state performance; at the same time, the calculation process of feedforward control is simple and clear, and the parameters are all obtained through offline experiments and clinical data, with high certainty and reliability; compared with the traditional method that solely relies on feedback regulation, this feedforward control scheme can compensate for system defects in advance before disturbances arrive, reducing the response time and control error required for feedback regulation, thereby greatly improving the accuracy and safety of anesthetic drug supply, providing a good initial benchmark for subsequent PID feedback control, making the entire control closed-loop smoother, faster, and more predictive, and effectively solving the problems of large initial errors and system response lags in the existing technology due to the lack of feedforward control.

[0154] Perform the following feedback correction:

[0155] Calculate the anesthetic drug concentration error:

[0156]

[0157] Where:

[0158] : is the anesthetic drug concentration in the patient's body obtained after filtering;

[0159] : is the target concentration set for the anesthetic drug in the patient's body;

[0160] Calculate the feedback correction amount:

[0161]

[0162] Where:

[0163] : is the feedback correction, representing the anesthetic drug infusion rate, unit: ;

[0164] : represents the drug concentration error;

[0165] : are the proportional, integral, and differential gains respectively, and the value range is , , ; and obtain the integral term from historical data and the differential term ;

[0166] Combine the feedforward control input and the feedback correction:

[0167]

[0168] Among them, is the infusion rate of the anesthetic drug input into the patient's body after correction. In the PID feedback control with feedforward compensation, by combining the reference obtained by feedforward control with the feedback correction amount calculated based on the current error , the overall control signal is formed; this design not only utilizes the steady-state reference provided by feedforward control but also instantaneously adjusts the real-time error through PID feedback, thereby achieving a fast response to dynamic disturbances and the elimination of steady-state errors; the proportional gain directly amplifies the current error, enabling the system to respond quickly; the integral term ensures the elimination of long-term deviations; the differential term estimates the trend of error change, which helps to suppress system overshoot and oscillation; through this combination of feedforward and feedback, a basic compensation can be made before the disturbance occurs, and at the same time, it can be quickly adjusted after the disturbance appears, making the response of the entire control system to errors smoother and more stable; this design not only overcomes the problems of integral saturation and overshoot that may occur in traditional PID control when responding to dynamic disturbances but also significantly improves its robustness and anti-interference ability because the feedback regulation is calculated based on real-time smooth data, thus making the anesthetic drug supply control more accurate, real-time, and safer, solving the problem of concentration fluctuations caused by lag response in existing single feedback control methods and significantly reducing the patient risk.

[0169] The optimization compensation according to the error state to obtain the infusion rate of the anesthetic drug input into the patient's body after compensation includes:

[0170] Establish a drug dynamic model:

[0171]

[0172] Among them: is the set target concentration of the anesthetic drug in the patient's body, equal to ; is the feedforward control output, equal to ;

[0173] Define the error:

[0174]

[0175] Linearize the perturbation to obtain the error dynamics:

[0176]

[0177] At introduce the deviation variables:

[0178]

[0179] Substituting gives:

[0180]

[0181] At steady state, , that is Finally, we get:

[0182]

[0183] Establish the state equation and let the state :

[0184]

[0185] Then we get , ;

[0186] Set the state weight and the control weight , and construct the cost function as:

[0187]

[0188] is the state weight, set artificially, the larger the value, the more strictly the error is punished;

[0189] is the control weight, set artificially, the larger the value, the smaller the change in the control input is desired;

[0190] Through the constructed cost function, the algebraic Riccati equation is obtained as:

[0191]

[0192] Solve the algebraic Riccati equation for the positive definite solution;

[0193] Calculate the optimal feedback gain :

[0194]

[0195] The optimal compensation control is as follows:

[0196]

[0197] Finally, the optimal control at the current moment is:

[0198]

[0199] wherein, is the infusion rate of the anesthetic drug finally determined to be input into the patient's body. The LQR method is used to perform optimal state feedback compensation on the system error. Its main advantage is that by constructing a quadratic cost function, the state error and the cost of the control input are comprehensively considered, so as to obtain an optimal feedback gain ; This method first linearizes and approximates the system pharmacokinetic model, and writes the error dynamics as a state space model , wherein represents the error between the target concentration and the actual concentration. This model fully reflects the dynamic characteristics of the system under small disturbance conditions; After constructing the cost function , a positive definite matrix is obtained by solving the algebraic Riccati equation, and then the optimal feedback gain is calculated; Using this feedback gain, the control compensation law can quickly zero the error while ensuring smooth changes in the control input; The beneficial effect of this method is that it not only provides a theoretically optimal compensation control increment, but also, since all parameters are determined by the physical model, it has high certainty and robustness; LQR compensation can effectively solve the dynamic response problem that is difficult to eliminate in traditional PID feedback control, significantly reducing the risk of system overshoot and oscillation, making the final control signal more accurate and stable; In addition, this method realizes the automatic adjustment of the control strength through quadratic cost function optimization, which helps to maintain the system performance under various interference conditions, thus greatly improving the safety and control accuracy during the supply process of anesthetic drugs, and effectively making up for the limitations existing when using PID feedback control or feedforward control alone in the prior art.

[0200] During the process of inputting anesthetic drugs into the patient's body, the data collected by the sensor is evaluated for abnormal scores, including:

[0201] During the control process, an abnormal score evaluation is carried out:

[0202]

[0203] wherein:

[0204] : time Abnormal score;

[0205] : The number of sensors for obtaining the patient's physiological parameters;

[0206] : The th sensor's measurement value at time ;

[0207] : The historical mean value of the th sensor;

[0208] : The standard deviation of the th sensor;

[0209] : The weights of each sensor, with the value range of ;

[0210] Set the abnormal determination threshold to ;

[0211] Compare the obtained at each moment with ;

[0212] If the at a certain moment, it is determined that there is an abnormality.

[0213] When the result of the abnormal score evaluation shows an abnormality, perform intelligent compensation, including:

[0214] When there is an abnormality, perform intelligent compensation, and the compensation amount is , and the calculation formula is:

[0215]

[0216] Where:

[0217] : Compensation adjustment amount;

[0218] : Sensitivity adjustment coefficient, with the value range of ;

[0219] : Hyperbolic tangent function, used to smooth the compensation output and prevent over-compensation.

[0220] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0221] The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A control method for a drug supply control system of an anesthesia machine for anesthesia department, characterized in that: include: The control system comprises: Multi-dimensional sensor module: used to collect the concentration of anesthetic drugs in the patient's body; Data acquisition and clock synchronization module: used to read sensor data in real time and record the sampling time to ensure the accuracy of the data timestamp; Data preprocessing module: Use weighted sliding average filtering technology to smooth the collected raw data and output smoothed data as input for subsequent control; Patient physiological parameter database: stores basic information and physiological indicators of patients, and provides individualized parameters for obtaining feedforward control output; Fault detection and fault tolerance module: monitors the data of each module in real time, detects anomalies through statistical methods, and determines whether to activate the fault tolerance mechanism; Drug dosage control module: used to receive drug dosage control signals and adjust the injection rate of anesthetic drugs into the patient's body according to the control signals; The control method comprises: During the anesthesia process of the patient, the concentration of anesthetic drugs in the patient's body is collected; Filter the collected patient anesthetic drug concentrations; According to the target concentration of the anesthetic drug in the patient's body, the basal drug infusion rate required to maintain the target concentration under steady-state conditions is calculated based on the pharmacokinetic model, thereby obtaining a feedforward control output that enables the anesthetic drug concentration in the patient's body to reach the target concentration; According to the target concentration and the filtered anesthetic drug concentration, dynamic disturbances are quickly compensated for by proportional, integral and differential adjustments for feedback correction; By constructing a quadratic cost function to calculate the state error and the control input cost, the optimal feedback gain is obtained, and then the compensation is optimized according to the error state to obtain the infusion rate of the anesthetic drug into the patient's body after compensation; During the process of delivering anesthetic drugs to the patient, the data collected by the sensor is evaluated for abnormality scores; When the result of the abnormality score assessment is abnormal, intelligent compensation is performed, including: When there is an abnormality, intelligent compensation is performed, and the compensation amount is , the calculation formula is: in: : compensation adjustment amount; : Sensitivity adjustment coefficient, the value range is ; : Hyperbolic tangent function; :time Abnormal score of : Abnormal determination threshold.

2. The control method of the drug supply control system of the anesthesia machine for anesthesia department according to claim 1, characterized in that: The filtering process of the collected patient anesthetic drug concentration includes: The collected anesthetic drug concentration in the patient's body is filtered, specifically: Will be at the time The anesthetic drug concentration collected from the patient is recorded as ; set up Sliding window of According to the set sliding window, the anesthetic drug concentrations of the last five unit times are collected and recorded as , , , , ; Set weights for the anesthetic drug concentration at each moment: ， ， ， ， in For the current moment The weight of For the last moment The weight of, and so on; Calculate the weighted moving average of the collected anesthetic drug concentrations in the patient: in: : Indicates at time The concentration of anesthetic drugs in the patient's body obtained after filtering; : Indicates at time The anesthetic drug concentration of the patient collected, ; : Indicates the preset weight of the anesthetic drug concentration at the corresponding moment.

3. The control method of the drug supply control system of the anesthesia machine for anesthesia department according to claim 1, characterized in that: The method of calculating the basal drug infusion rate required to maintain the target concentration under steady-state conditions based on the pharmacokinetic model according to the target concentration of the anesthetic drug in the patient's body, and then obtaining the feedforward control output that enables the anesthetic drug concentration in the patient's body to reach the target concentration, includes: Set the target concentration of anesthetic drugs in the patient's body, denoted as ; When the concentration of anesthetic drugs in the patient's body is required to be When the anesthetic drug is injected into the patient's body, the infusion rate is: in: : Distribution volume, determined according to the patient's weight, with a range of ; : Drug elimination rate constant, unit , the value range is ; : Feedforward control output, indicating the anesthetic drug infusion rate, unit: .

4. The control method of the drug supply control system of the anesthesia machine for anesthesia department according to claim 1, characterized in that: The feedback correction is performed by quickly compensating for dynamic disturbances through proportional, integral and differential regulation: Calculate the error in anesthetic drug concentration: in: : is the concentration of anesthetic drugs in the patient’s body obtained after filtering; : The target concentration of anesthetic drugs set for the patient; Calculate the feedback correction: in: : is the feedback correction, indicating the infusion rate of anesthetic drugs, unit: ; : Indicates drug concentration error; : They are proportional, integral and differential gains respectively, and their value range is , , ; and obtain the integral item from historical data and the differential term ; Combine the feedforward control input with the feedback correction: in, Indicates the feedforward control output, indicating the anesthetic drug infusion rate, unit: , The corrected infusion rate of anesthetic drugs into the patient.

5. The control method of the drug supply control system of the anesthesia machine for anesthesia department according to claim 1, characterized in that: The method calculates the state error and the control input cost by constructing a quadratic cost function to obtain the optimal feedback gain, thereby optimizing compensation according to the error state and obtaining the infusion rate of the anesthetic drug into the patient after compensation, including: Establish drug dynamic model: in: is the target concentration of anesthetic drugs in the patient's body, equal to ; is the feedforward control output, equal to ; : Drug elimination rate constant, unit , the value range is ; : Distribution volume, determined according to the patient's weight, with a range of ; Define the error: Linearizing the disturbance yields the error dynamics: exist When , the deviation variable is introduced: Substituting in, we get: In steady state, ,Right now , and finally get: Establish the state equation and let the state : Then we get , ; Setting state weight and control weights , and construct the cost function as: Through the constructed cost function, the algebraic Riccati equation is obtained as: Solving the algebraic Riccati equation yields The positive solution of ; Calculate the optimal feedback gain : The optimal compensation control for: Finally, the optimal control at the current moment is: in, The final determined infusion rate of anesthetic drugs into the patient's body.

6. The control method of the drug supply control system of the anesthesia machine for anesthesia department according to claim 1, characterized in that: During the process of injecting anesthetic drugs into the patient's body, the data collected by the sensor is evaluated for abnormality scores, including: During the control process, abnormal score evaluation is performed: in: :time Abnormal score of : The number of sensors that obtain the patient's physiological parameters; : No. The sensor at the time The measured value of : No. The historical average of the sensors; : No. The standard deviation of each sensor; : The weight of each sensor, the value range is ; Set the abnormality judgment threshold to ; The and Make comparisons; If at some point , it is determined that there is an abnormality.

Images