Global intelligent optimal control method applied to flushing, perfusion and heating system

Abstract

The invention provides a global intelligent optimal control method applied to an irrigation and perfusion heating system, and relates to the technical field of data processing. According to clinical element information analysis, a safe irrigation and perfusion temperature threshold and a safe irrigation and perfusion flow rate threshold are obtained for medical staff to select and set in real time; and optimizing and outputting a multi-stage heating parameter sequence by taking energy consumption minimization as a target, monitoring a result congestion risk in a flushing, filling and heating control process by adopting the multi-stage heating parameter sequence, and activating a redundancy switching mechanism to switch to a standby water outlet when the congestion risk exists. The technical problem that in the prior art, parameter setting of a flushing, perfusion and heating system depends on artificial experience, management and control in the use process are lacked, and use safety and continuity are affected is solved. The technical effect that the energy efficiency level and the system robustness of the flushing, filling and heating system are remarkably improved on the premise that the use safety and continuity of the flushing, filling and heating system are guaranteed is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a global intelligent optimal control method applied to a flushing perfusion warming system. BACKGROUND

[0002] The technical defects of the existing flushing perfusion warming system are caused by the disconnection between clinical decision and device execution. Specifically, parameter setting relies on the experience of medical staff, without integrating operation type, patient individual differences and device physical limits, resulting in frequent non-adaptive settings.

[0003] At the same time, the device execution stage lacks early sensing ability for water flow blockage signs, and can only be passively alarmed when the water flow is completely interrupted, which forces the interruption of device use, seriously threatening the safety and continuity of device use.

[0004] In summary, the existing technology has the technical problem that the parameter setting of the flushing perfusion warming system relies on manual experience and lacks process control during use, which affects the safety and continuity of use. SUMMARY

[0005] In view of the above defects or improvement needs of the prior art, the present application provides a global intelligent optimal control method applied to a flushing perfusion warming system, which is used to solve the technical problem that the parameter setting of the flushing perfusion warming system relies on manual experience and lacks process control during use in the prior art, which affects the safety and continuity of use.

[0006] To achieve the above-mentioned purpose, the present application provides a global intelligent optimal control method applied to a flushing perfusion warming system, which comprises:

[0007] After the flushing perfusion warming system enters the operation support mode, the clinical element information is extracted from the received operation reservation instruction. After verifying the clinical element information, the clinical adaptive safety boundary coupling analysis is performed according to the clinical element information, and the safe flushing perfusion temperature threshold and the safe flushing perfusion flow rate threshold are obtained. After the safe flushing perfusion temperature threshold and the safe flushing perfusion flow rate threshold are displayed on the system operation interface, the real-time set temperature and the real-time set flow rate returned by the medical staff interacting with the system operation interface are received. With the goal of minimizing energy consumption, the thermal management control parameter optimization is performed according to the real-time set temperature and the real-time set flow rate, and the multi-stage heating parameter sequence is output. During the execution of the flushing perfusion warming control process using the multi-stage heating parameter sequence, the water flow fluctuation of the main water outlet of the flushing perfusion warming system is monitored, and the congestion risk prediction is performed according to the monitoring result. If there is a congestion risk in the main water outlet, the redundant switching mechanism is activated to switch the water outlet task to the backup water outlet.

[0008] In one embodiment, the following processing is also performed:

[0009] The irrigation perfusion warming system receives a surgery appointment instruction, parses a surgery start timestamp in the surgery appointment instruction, obtains a mode switching node of the surgery support mode, takes the mode switching node as a reference, reversely traces a dynamic self-check time window, locates a device self-check time node, performs multi-dimensional component detection on the irrigation perfusion warming system at the device self-check time node, outputs a device self-check report, and switches the irrigation perfusion warming system to the surgery support mode at the mode switching node only when the report conclusion of the device self-check report is full-dimensionally qualified.

[0010] In an embodiment, after verifying the clinical element information, a clinical adaptation safety boundary coupling analysis is performed according to the clinical element information to obtain a safe irrigation perfusion temperature threshold and a safe irrigation perfusion flow rate threshold, and the following processing is further performed:

[0011] A clinical element vector is constructed based on the clinical element information, a plurality of element similarities of the clinical element vector and a plurality of historical element vectors in a historical irrigation control library are calculated, P historical element vectors are screened based on a preset similarity threshold, P historical irrigation perfusion temperatures and P historical irrigation perfusion flow rates of the P historical element vectors are called, the P historical irrigation perfusion temperatures and the P historical irrigation perfusion flow rates are weighted and averaged to aggregate the P historical irrigation perfusion temperatures and the P historical irrigation perfusion flow rates according to a set recurrence frequency of the P historical irrigation perfusion temperatures and the P historical irrigation perfusion flow rates to obtain an initial irrigation perfusion temperature threshold and an initial irrigation perfusion flow rate threshold, a device safety boundary database is invoked, and the initial irrigation perfusion temperature threshold and the initial irrigation perfusion flow rate threshold are corrected by using the device safety boundary database to obtain the safe irrigation perfusion temperature threshold and the safe irrigation perfusion flow rate threshold.

[0012] In an embodiment, the clinical element information includes a surgery type code, a patient body temperature floating point value, an irrigation fluid storage temperature value, an irrigation fluid pH value, a patient age segment, and an irrigation fluid viscosity coefficient.

[0013] In an embodiment, a thermal management control parameter optimization is performed according to the real-time set temperature and the real-time set flow rate to minimize energy consumption, a multi-stage heating parameter sequence is output, and the following processing is further performed:

[0014] The real-time set temperature and real-time set flow rate are taken as joint search conditions to perform initial population search to obtain multiple sample heating parameter settings, wherein the sample heating parameter settings include sample power gear combination sequences, sample power gear time allocation, and sample cavity airflow circulation parameters; energy consumption calculation is performed on the multiple sample heating parameter settings to obtain multiple sample energy consumption requirements; a preset energy consumption baseline is used to traverse the multiple sample energy consumption requirements to filter W reference heating parameter settings from the multiple sample heating parameter settings; after a heating parameter setting space is constructed according to indexes of the W reference heating parameter settings, an optimization envelope space is framed based on the W reference heating parameters in the heating parameter setting space; parameter combination iterative search and energy consumption calculation are performed in the optimization envelope space until a convergence condition is reached, and the multiple-stage heating parameter sequence is output.

[0015] In an embodiment, the following processing is also performed:

[0016] During execution of the flushing perfusion warming control process using the multiple-stage heating parameter sequence, the real-time outlet water temperature and real-time outlet water flow rate of the main outlet are monitored; flow pump dynamic PID fine tuning is performed according to the real-time outlet water flow rate and flow rate deviation of the real-time outlet water flow rate; heating power dynamic PID fine tuning is performed according to the real-time outlet water temperature and temperature deviation of the real-time set temperature, and when the temperature deviation exceeds a preset safety tolerance, the flow pump dynamic PID fine tuning enters an output frozen state; wherein the heating power dynamic PID fine tuning and the flow pump dynamic PID fine tuning share deviation threshold trigger rules.

[0017] In an embodiment, during execution of the flushing perfusion warming control process using the multiple-stage heating parameter sequence, water flow fluctuation monitoring is performed on the main outlet of the flushing perfusion warming system, and congestion risk prediction is performed according to the monitoring result, and the following processing is also performed:

[0018] Data backtracking of historical congestion events is performed with a congestion risk window as a constraint, multiple sample time series flow rate drop rates, multiple sample time series pressure rise slopes, multiple sample time series fluctuation frequencies, and multiple sample event congestion probability values; a congestion risk prediction model is constructed based on an LSTM network; the multiple sample time series flow rate drop rates, the multiple sample time series pressure rise slopes, the multiple sample time series fluctuation frequencies, and the multiple sample event congestion probability values are taken as training data to perform parameter adjustment optimization of the congestion risk prediction model.

[0019] In an embodiment, during execution of the flushing perfusion warming control process using the multiple-stage heating parameter sequence, water flow fluctuation monitoring is performed on the main outlet of the flushing perfusion warming system, and congestion risk prediction is performed according to the monitoring result, and the following processing is also performed:

[0020] The congestion risk window performs water flow fluctuation monitoring of the main water outlet in a predefined sliding scale, obtaining a plurality of real-time time series flow rate drop rates, a plurality of real-time time series pressure rise slopes, and a plurality of real-time time series fluctuation frequencies; the plurality of real-time time series flow rate drop rates, the plurality of real-time time series pressure rise slopes, and the plurality of real-time time series fluctuation frequencies are intermittently loaded into the congestion risk prediction model based on the predefined sliding scale, congestion risk prediction is performed, and a plurality of real-time event congestion probabilities are output; when any of the plurality of real-time event congestion probabilities satisfies a preset congestion threshold, a congestion event early warning is output; based on the congestion event early warning, a redundancy switching mechanism is activated, and the water outlet task is switched to the backup water outlet.

[0021] The technical solutions provided in the present application have at least the following technical effects or advantages:

[0022] The method provided by the embodiment of the present application extracts clinical element information from receiving a surgical appointment instruction after the flushing perfusion warming system enters the surgical support mode; after verifying the clinical element information, the clinical element information is coupled for clinical adaptation safety boundary analysis, to obtain a safe flushing perfusion temperature threshold and a safe flushing perfusion flow rate threshold; after the safe flushing perfusion temperature threshold and the safe flushing perfusion flow rate threshold are displayed to the system operation interface, real-time set temperature and real-time set flow rate returned by medical staff interacting with the system operation interface are received; with the goal of minimizing energy consumption, the real-time set temperature and the real-time set flow rate are used to perform thermal management control parameter optimization, and a multi-stage heating parameter sequence is output; during the flushing perfusion warming control process using the multi-stage heating parameter sequence, the main water outlet of the flushing perfusion warming system is monitored for water flow fluctuation, and congestion risk prediction is performed according to the monitoring result; if the main water outlet has a congestion risk, a redundancy switching mechanism is activated to switch the water outlet task to the backup water outlet. The technical problem of affecting the safety and continuity of use due to the dependence of parameter setting of the flushing perfusion warming system on manual experience and the lack of process control in the prior art is solved. The technical effect of significantly improving the energy efficiency and system robustness of the flushing perfusion warming system is achieved under the premise of ensuring the safety and continuity of use of the flushing perfusion warming system. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the following embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0024] Figure 1 A global intelligent optimization method flowchart applied to a flushing perfusion warming system provided by the present application is shown.

[0025] Figure 2 The application provides a flowchart for acquiring a safe flushing perfusion temperature threshold and a safe flushing perfusion flow rate threshold in the global intelligent optimal control method applied to the flushing perfusion warming system. DETAILED DESCRIPTION

[0026] The application provides a global intelligent optimal control method applied to a flushing perfusion warming system, which is used to solve the technical problem that the parameter setting of the flushing perfusion warming system in the prior art depends on artificial experience and lacks process control in use, thereby affecting the safety and continuity in use.

[0027] In order to make the objectives, technical solutions and advantages of the present application clearer, the following further describes the present application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application. In addition, the technical features involved in the various embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0028] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise" and the like indicate the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the present application and simplify the description, and therefore cannot be understood as indicating or implying that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0029] Unless otherwise explicitly indicated, throughout the specification and claims, the term "comprise" or its variants such as "contain" or "include" and the like will be understood to include the stated element or component, but not exclude the presence of other elements or components.

[0030] Embodiment, the flowchart of the global intelligent optimal control method applied to the flushing perfusion warming system provided by the embodiment of the present application, see Figure 1 , the method comprises:

[0031] Step A100: After the flushing perfusion warming system enters the operation support mode, extracting clinical element information from the received operation appointment instruction.

[0032] In one implementation, the clinical element information includes an operation type code, a patient temperature floating point value, a flushing fluid preservation temperature value, a flushing fluid pH value, a patient age segment, and a flushing fluid viscosity coefficient.

[0033] It should be understood that the operation appointment instruction is a structured digital instruction issued by a hospital information system (HIS) to the irrigation perfusion warming system, which carries the operation type, patient identification, appointment timestamp, and associated clinical parameters (such as body temperature and irrigation fluid properties) to trigger the irrigation perfusion warming system to start the clinical element extraction and equipment self-checking process.

[0034] The irrigation perfusion warming system first performs the equipment self-checking process after receiving the operation appointment instruction, and only when the self-checking is qualified, the irrigation perfusion warming system enters the operation support mode.

[0035] Specifically, after the irrigation perfusion warming system enters the operation support mode, the irrigation perfusion warming system extracts multi-dimensional clinical element information as individualized decision basis by analyzing the received operation appointment instruction.

[0036] These clinical element information includes standardized coding identifying the operation type (such as orthopedic debridement or abdominal perfusion), basic body temperature floating point value reflecting the physiological state of the patient, preservation temperature value of the irrigation fluid in the storage environment, pH value affecting the thermal stability of the fluid, age segment (such as children / elderly / adults) associated with the metabolic characteristics of the patient, and viscosity coefficient (such as high-viscosity thrombin solution) representing the flow characteristics of the fluid.

[0037] These elements in the clinical element information collectively constitute the input basis for clinical adaptation safety boundary analysis, providing multi-source clinical decision support for subsequent generation of safety temperature / flow rate threshold by fusing operation requirements, patient individual differences, and fluid physical property parameters.

[0038] Step A200: After verifying the clinical element information, a clinical adaptation safety boundary coupling analysis is performed according to the clinical element information to obtain a safe irrigation perfusion temperature threshold and a safe irrigation perfusion flow rate threshold.

[0039] In one implementation, Figure 2 It is shown that, after verifying the clinical element information, a clinical adaptation safety boundary coupling analysis is performed according to the clinical element information to obtain a safe irrigation perfusion temperature threshold and a safe irrigation perfusion flow rate threshold, the method step A200 provided by the present application comprises:

[0040] Step A210: Constructing a clinical element vector based on the clinical element information.

[0041] Step A220: After calculating the similarity of the clinical element vector and multiple historical element vectors in the historical irrigation control library, filtering P historical element vectors based on a preset similarity threshold.

[0042] Step A230: Calling P historical irrigation perfusion temperatures and P historical irrigation perfusion flow rates of the P historical element vectors.

[0043] Step A240: According to the setting recurrence frequency of the P historical irrigation perfusion temperatures and P historical irrigation perfusion flow rates, the P historical irrigation perfusion temperatures and P historical irrigation perfusion flow rates are weighted and averaged to obtain an initial irrigation perfusion temperature threshold and an initial irrigation perfusion flow rate threshold.

[0044] Step A250: The device safety boundary database is called and used to constrain and correct the initial irrigation perfusion temperature threshold and the initial irrigation perfusion flow rate threshold to obtain the safety irrigation perfusion temperature threshold and the safety irrigation perfusion flow rate threshold.

[0045] Specifically, the surgical type code, the patient temperature floating point value, the irrigation fluid preservation temperature value, the irrigation fluid pH value, the patient age segment, and the irrigation fluid viscosity coefficient contained in the clinical element information are coded into a structured multi-dimensional feature vector to obtain the clinical element vector. The clinical element vector integrates discrete data and provides standardized input for historical data matching, ensuring the comparability of features in different surgical scenarios.

[0046] The historical database stores a plurality of historical element vectors corresponding to a plurality of historical clinical element information preprocessed by structured coding, and a plurality of historical irrigation perfusion temperature-historical irrigation perfusion flow rate data groups corresponding to the plurality of historical clinical element information, and the plurality of historical element vectors and the plurality of historical irrigation perfusion temperature-historical irrigation perfusion flow rate data groups are stored in association.

[0047] After calculating the element similarity between the clinical element vector and a plurality of historical element vectors in the historical irrigation control library using a non-specific similarity calculation method such as Euclidean distance or cosine similarity, P historical element vectors of P high-similarity historical cases are selected based on a preset similarity threshold (such as >0.85) to locate similar surgical scenarios through data-driven and inherit historical parameter setting experience.

[0048] The P historical irrigation perfusion temperatures and P historical irrigation perfusion flow rates of the P historical element vectors are called, and the P parameters represent feasible solutions repeatedly verified in similar surgeries.

[0049] According to the setting recurrence frequency statistical result of the P historical irrigation perfusion temperatures and P historical irrigation perfusion flow rates, P frequency weight coefficients are set, and the P historical irrigation perfusion temperatures and P historical irrigation perfusion flow rates are weighted and averaged according to the P frequency weight coefficients to realize that high-frequency recurring parameters dominate the aggregation result and low-frequency parameters have attenuated influence, thereby obtaining the final initial irrigation perfusion temperature threshold and the initial irrigation perfusion flow rate threshold, and achieving the technical effect of converting discrete historical experience into continuous and usable clinical decision benchmarks.

[0050] The device safety boundary database (containing heating power upper limit, pipeline pressure limit, etc.) is called to correct the initial threshold value with physical constraints as rigid conditions: if the initial threshold value exceeds the device capacity (such as the initial threshold value 43℃ but the device upper limit 42℃), it is truncated to the device limit value, and the final output includes the safety threshold interval of the safety flush perfusion temperature threshold and the safety flush perfusion flow rate threshold, which ensures that the obtained safety flush perfusion temperature threshold and safety flush perfusion flow rate threshold are compatible with clinical needs and device reliability.

[0051] Step A300: After displaying the safety flush perfusion temperature threshold and the safety flush perfusion flow rate threshold to the system operation interface, the real-time set temperature and the real-time set flow rate returned by the medical staff interacting with the system operation interface are received.

[0052] Specifically, the safety flush perfusion temperature threshold and the flow rate threshold calculated in this embodiment are pushed to the operation interface in the form of visual interval limits (for example, the safety boundary values at both ends of the temperature setting sliding bar), and the medical staff manually sets the specific temperature / flow rate value within the safety threshold range according to the actual operation demand. The flush perfusion warming system receives the real-time set parameters (real-time set temperature and real-time set flow rate) confirmed by the person as the subsequent control reference.

[0053] It should be noted that this step converts the intelligent analysis result into decision assistance through human-computer interaction design, the core function is the optimization configuration of device control parameters, and it does not involve human physiological diagnosis or health condition evaluation. The operation object is the medical device itself, and the set parameters are used for optimizing the medical device parameter setting process rather than obtaining a diagnostic conclusion.

[0054] Step A400: In order to minimize energy consumption, the heat management control parameter optimization is performed according to the real-time set temperature and the real-time set flow rate, and a multi-stage heating parameter sequence is output.

[0055] In one implementation, in order to minimize energy consumption, the heat management control parameter optimization is performed according to the real-time set temperature and the real-time set flow rate, and a multi-stage heating parameter sequence is output. The method step A400 provided by the present application includes:

[0056] Step A410: The real-time set temperature and the real-time set flow rate are used as joint search conditions to perform initial population search, and a plurality of sample heating parameter settings are obtained, wherein the sample heating parameter setting includes a sample power gear combination sequence, a sample power gear time allocation, and a sample cavity airflow circulation parameter.

[0057] Step A420: Energy consumption calculation is performed on the plurality of sample heating parameter settings to obtain a plurality of sample energy consumption requirements.

[0058] Step A430: Traverse the plurality of sample energy consumption requirements using a preset energy consumption baseline to filter W reference heating parameter settings from the plurality of sample heating parameter settings.

[0059] Step A440: According to the index composition of the W reference heating parameter settings, construct a heating parameter setting space, and then frame an optimization envelope space based on the W reference heating parameters in the heating parameter setting space.

[0060] Step A450: Perform parameter combination iterative search and energy consumption calculation in the optimization envelope space until the convergence condition is reached, and output the multi-stage heating parameter sequence.

[0061] Specifically, the real-time set temperature and real-time set flow rate are taken as joint search conditions to perform initial population search in the historical operation database, and a plurality of historical sample heating parameter settings with similar set requirements are matched to obtain a plurality of historical sample heating parameter settings, which include time sequence combination of power gears (such as “high gear-medium gear-low gear” sequence), accurate duration of each gear, and configuration parameters of air flow speed and circulation mode. These samples are derived from verified energy-saving control practices in similar device application scenarios.

[0062] Based on the physical characteristics of the device (power-gear mapping relationship of the heating module, speed-power consumption cubic curve of the air flow fan), theoretical total energy consumption calculation is performed on each sample parameter setting. The calculation model focuses on the energy efficiency characteristics of the device itself and does not involve human physiological parameters or health status analysis. It is a pure mathematical modeling process of industrial equipment power consumption, and a plurality of sample energy consumption requirements are obtained.

[0063] A preset energy consumption baseline value (such as 1.2 times the historical average energy consumption of the operation type) is used as a screening threshold to traverse the energy consumption calculation results of all samples, and only low-power samples with energy consumption requirements below the threshold are retained to form W reference heating parameter settings.

[0064] According to the index composition of the W reference heating parameter settings, construct a heating parameter setting space, and then abstract the power gear switching rule, heating time distribution interval, and air flow parameter optional set of the W reference parameter settings to construct a mathematical feasible region space (i.e., an optimization envelope space) of device control parameters. The optimization envelope space frames the boundary of the optimization search, ensuring that the newly generated parameter combination not only continues the historical energy-saving experience, but also has innovation exploration flexibility.

[0065] Performing random parameter combination search within the optimization envelope space: generating candidate control scheme → calculating energy consumption and checking temperature / flow rate stability → retaining high-quality solutions that meet safety constraints. Iterating until the scheme energy consumption converges (e.g., the optimization amplitude is less than 1% for 20 consecutive iterations), outputting a time-axis-encoded multi-level heating parameter sequence (e.g., "0-3 minutes: high gear + high-speed airflow. 3-15 minutes: medium gear + circulating airflow").

[0066] The embodiment achieves the technical effect of systematically reducing the total energy consumption of the equipment while ensuring the control accuracy of the flushing temperature / flow rate.

[0067] Step A500: During the execution of the flushing perfusion warming control process using the multi-level heating parameter sequence, the water flow fluctuation of the main water outlet of the flushing perfusion warming system is monitored, and the congestion risk prediction is performed according to the monitoring result.

[0068] In one implementation, during the execution of the flushing perfusion warming control process using the multi-level heating parameter sequence, the water flow fluctuation of the main water outlet of the flushing perfusion warming system is monitored, and the congestion risk prediction is performed according to the monitoring result. Before that, the method step A500 provided by the present application includes:

[0069] Step A5001: Data backtracking of historical congestion events is performed with congestion risk window as constraint, multiple sample time series flow rate drop rates, multiple sample time series pressure rise slopes, multiple sample time series fluctuation frequencies, and multiple sample event congestion probability values.

[0070] Step A5002: A congestion risk prediction model is constructed based on an LSTM network.

[0071] Step A5003: The multiple sample time series flow rate drop rates, multiple sample time series pressure rise slopes, multiple sample time series fluctuation frequencies, and multiple sample event congestion probability values are used as training data to perform parameter tuning optimization of the congestion risk prediction model.

[0072] In one implementation, during the execution of the flushing perfusion warming control process using the multi-level heating parameter sequence, the water flow fluctuation of the main water outlet of the flushing perfusion warming system is monitored, and the congestion risk prediction is performed according to the monitoring result. The method step A500 provided by the present application includes:

[0073] Step A510: The congestion risk window performs water flow fluctuation monitoring of the main water outlet with a pre-defined sliding scale, obtaining multiple real-time time series flow rate drop rates, multiple real-time time series pressure rise slopes, and multiple real-time time series fluctuation frequencies.

[0074] Step A520: The plurality of real-time time-series flow rate drop rates, the plurality of real-time time-series pressure rise slopes, and the plurality of real-time time-series fluctuation frequencies are intermittently loaded into the congestion risk prediction model based on the predefined sliding scale, a congestion risk prediction is performed, and a plurality of real-time event congestion probabilities are output.

[0075] Step A530: When any of the plurality of real-time event congestion probabilities satisfies a preset congestion threshold, a congestion event warning reminder is output.

[0076] Step A540: Based on the congestion event warning reminder, a redundancy switching mechanism is activated, and a water outlet task is switched to the backup water outlet.

[0077] Specifically, in the present embodiment, a pre-constructed congestion risk prediction model is used to predict the congestion risk of the main water outlet of the flushing perfusion warming system.

[0078] The construction process of the congestion risk prediction model is as follows:

[0079] With the congestion risk time window as a constraint condition (such as 60 seconds before the occurrence of congestion), the recorded typical congestion cases are retrieved from the historical congestion failure database of the flushing perfusion warming system. The complete time-series data of each case in the risk window is extracted: including the flow rate drop rate change curve (flow rate reduction percentage per unit time), the pressure rise slope trend (kPa / s), and the water flow fluctuation frequency spectrum (0.1-5 Hz energy distribution), and the corresponding real congestion probability value (occurring congestion = 1.0, not occurring = 0) is labeled. The present embodiment performs data retrieval to construct a structured training sample set, which provides time-series features covering various congestion scenarios for the congestion risk prediction model.

[0080] The specific data composition of the structured training sample set includes: a plurality of sample time-series flow rate drop rates, a plurality of sample time-series pressure rise slopes, a plurality of sample time-series fluctuation frequencies, and a plurality of sample event congestion probability values.

[0081] Based on the long short-term memory network (LSTM), a time-series prediction framework of the congestion risk prediction model is built. The network input layer is designed to have a dimension of 60x4 (60 time steps x flow rate / pressure / fluctuation frequency / temperature compensation four features), the hidden layer contains 128 LSTM units to capture long-range dependencies, and the output layer uses a Sigmoid activation function to generate congestion probability values of 0-1.0. The network structure is optimized for extracting time-series patterns of water flow dynamics, such as the congestion feature pattern of sudden pressure rise accompanied by sudden flow rate drop.

[0082] In the congestion risk prediction model training and parameter optimization process, the time series flow rate drop rate, pressure rise slope and fluctuation frequency characteristics obtained by historical backtracking are taken as input, and the corresponding congestion probability label is taken as supervision signal. The cross-entropy loss of the predicted value and the true label is calculated by forward propagation, the LSTM network weight parameters are updated by gradient descent algorithm, and the learning rate and dropout rate and other hyperparameters are dynamically adjusted combined with the performance of the validation set. The training process embeds the device-specific bias, which applies three times the loss weight to the false alarm event to prioritize clinical safety, and uses the early stopping mechanism to terminate the training when the recall rate of the validation set is greater than 90% and the false alarm rate is less than 5%, and finally generates a lightweight congestion risk prediction model (<5MB) that meets the embedded device deployment requirements.

[0083] Specifically, in the process of controlling the flushing perfusion warming system to perform flushing perfusion warming control by using the multi-stage heating parameter sequence, the flow rate, pressure, and fluctuation frequency original signals of the main water outlet are continuously collected in a preset time window (e.g., every 10 seconds as a monitoring period). After sliding average filtering and trend decomposition processing, the standardized time series features are output, including the flow rate drop rate change value per unit time window, the pressure rise slope cumulative value, and the feature frequency band fluctuation energy proportion. These features are completely aligned with the structured definition in the training phase, ensuring the consistency of the model input.

[0084] The real-time extracted time series features are batched and input into the pre-trained congestion risk prediction model according to the monitoring period. The congestion risk prediction model outputs the congestion probability value (0.0-1.0 continuous quantity) of the current time window based on the mapping relationship between water flow dynamics and historical congestion events. The prediction process is executed every 30 seconds, balancing real-time performance and accuracy under the constraint of computing resources.

[0085] When the congestion probability output by any monitoring period exceeds the preset threshold (e.g., >0.85), a congestion event warning is immediately generated. The preset congestion threshold is determined through experimental testing to ensure that a high-probability warning corresponds to a >95% real congestion risk, minimizing the interference of false alarms on the flushing perfusion warming system application scenario usage process.

[0086] Based on the warning signal, a redundant switching mechanism is automatically activated: the standby pipeline system takes over the water outlet task within 500ms, and the main water outlet enters a maintenance state. The switching process uses an S-shaped flow transition curve to ensure that the water temperature fluctuation is <0.3°C and the flow rate change is <5%, avoiding sudden changes in water supply pressure parameters affecting surgical operations.

[0087] This embodiment captures water flow dynamic characteristics in real time and loads a pre-trained congestion prediction model to predict pipeline congestion risk in advance, providing a decision basis for redundant switching and ensuring the continuity of the flushing perfusion warming system usage.

[0088] Step A600: If the main water outlet is at risk of congestion, activate the redundancy switching mechanism to switch the water outlet task to the backup water outlet.

[0089] Specifically, after monitoring that the main water outlet is at risk of congestion, the embodiment activates the redundancy switching mechanism to switch the water outlet task to the backup water outlet, achieving the technical effect of significantly improving the energy efficiency level and system robustness of the irrigation perfusion warming system under the premise of ensuring the use safety and continuity of the irrigation perfusion warming system.

[0090] In an implementation manner, the method further comprises:

[0091] Step A1001: After receiving the operation appointment instruction, the irrigation perfusion warming system parses the operation start timestamp in the operation appointment instruction to obtain the mode switching node of the operation support mode.

[0092] Step A1002: Taking the mode switching node as a reference, the dynamic self-checking time window is traced back reversely to locate the equipment self-checking time node.

[0093] Step A1003: At the equipment self-checking time node, multi-dimensional component detection is performed on the irrigation perfusion warming system, and an equipment self-checking report is output.

[0094] Step A1004: When and only when the report conclusion of the equipment self-checking report is full-dimensionally qualified, the irrigation perfusion warming system is switched to the operation support mode at the mode switching node.

[0095] Specifically, after receiving the operation appointment instruction, the embodiment parses the operation start timestamp (such as "2024-06-25 09:30:00") carried in the instruction to accurately lock the activation time of the operation support mode. The operation start timestamp serves as a reference node for equipment state conversion (mode switching) and provides a time sequence anchor point for subsequent self-checking scheduling.

[0096] Taking the mode switching node as a reference, the dynamic self-checking time window (such as 2 hours before the operation) is reversely calculated. The dynamic self-checking time window is intelligently adjusted according to the operation type: the window length is fixed for regular operations, and the window length is compressed according to the emergency degree (minimum 30 minutes) for emergency operations. The optimal self-checking time node (such as avoiding the peak of hospital electricity consumption) is located within the dynamic self-checking time window to ensure that the self-checking process does not affect the normal operation of the operating room.

[0097] At the dynamically determined device self-check time node, start the full-link closed-loop verification of the flushing perfusion warming system: through infrared thermal imaging to scan the heating module surface temperature distribution uniformity (temperature difference tolerance ± 0.5℃), use a standard source to perform three-point method value traceability calibration on the temperature sensor (error ≤±0.1℃), drive the flow pump to run in idle state and capture the speed fluctuation rate (threshold value <2%), at the same time, apply negative pressure test to the sterile pipeline to test the sealing performance (pressure drop rate <0.5kPa / min), and verify the validity period and sterilization batch number compliance of the disposable flushing perfusion warming box by RFID radio frequency technology. All test results are fused by algorithm to generate a structured device self-check report output.

[0098] When and only when the report conclusion of the device self-check report is full-dimensionally qualified, at the mode switching node, switch the flushing perfusion warming system to the surgery support mode.

[0099] The self-check access mechanism of the embodiment realizes the technical effect of eliminating the subjective blind area of traditional manual inspection and ensuring the underlying reliability of the flushing perfusion warming system by dual data-driven decisions of hardware performance and consumable compliance.

[0100] Step A710: During the flushing perfusion warming control process performed by using the multi-level heating parameter sequence, monitor the real-time outlet water temperature and real-time outlet water flow rate of the main outlet.

[0101] Step A720: According to the real-time outlet water flow rate and flow rate deviation of the real-time outlet water flow rate, perform flow pump dynamic PID fine tuning.

[0102] Step A730: According to the temperature deviation of the real-time outlet water temperature and real-time set temperature, perform heating power dynamic PID fine tuning, and when the temperature deviation exceeds the preset safety tolerance, the flow pump dynamic PID fine tuning enters the output frozen state.

[0103] Wherein, the heating power dynamic PID fine tuning and the flow pump dynamic PID fine tuning share the deviation threshold trigger rule.

[0104] Specifically, during the execution of the multi-level heating parameter sequence, the real-time outlet water temperature (accuracy ±0.1℃) and real-time outlet water flow rate (accuracy ±1.5%) of the main outlet are synchronously collected, and the monitoring mechanism provides a data basis for subsequent dynamic fine tuning.

[0105] According to the deviation value of the real-time water outlet flow rate and the set flow rate (such as set 2L / min, measured 1.8L / min→deviation 10%), the flow pump dynamic PID fine tuning is automatically executed: through the proportional coefficient to accelerate the approach to the target, the integral term to eliminate the steady-state error, and the differential to suppress overshoot, the flow rate is stabilized by dynamically adjusting the pump speed (such as speed increase by 8%), and the flow pump dynamic PID fine tuning is strictly limited within the safe speed range of the device (such as ±15%), so as to avoid over-limit damage to the pump body.

[0106] In the temperature priority type PID cooperative control, the heating power is dynamically adjusted according to the deviation of the real-time outlet water temperature and the set temperature: when the temperature deviation is within the safe tolerance (such as ≤±0.5℃), the heating power PID fine tunes the power output through the proportional-integral-derivative algorithm (for example, when the deviation is +0.3℃, the power is increased by 3%). Once the temperature deviation exceeds the safe tolerance (such as >±0.5℃), the speed regulation instruction of the flow pump PID is immediately frozen (forced to maintain the current flow rate unchanged), and the control resources are concentrated to eliminate the temperature anomaly first, and after the temperature returns to the safe interval, the flow rate PID is automatically unfrozen. In this mechanism, the temperature and flow rate PID share the same deviation trigger threshold (such as ±5% deviation to start adjustment).

[0107] The embodiment achieves the technical effect of rigidly blocking the risk of thermal runaway through the temperature priority arbitration mechanism under the premise of ensuring the stability of the flushing perfusion parameters, and realizing the gain cooperation of double-loop control relying on the shared deviation threshold.

[0108] The above is only a preferred embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A global intelligent optimization control method applied to a flushing and heating system, characterized in that, include: After the irrigation and heating system enters the surgical support mode, it extracts clinical element information from the received surgical appointment instruction; After verifying the clinical element information, a clinical adaptation safety boundary coupling analysis was performed based on the clinical element information to obtain the safe flushing and perfusion temperature threshold and the safe flushing and perfusion flow rate threshold. After displaying the safe flushing and irrigation temperature threshold and the safe flushing and irrigation flow rate threshold on the system operation interface, the system receives the real-time set temperature and real-time set flow rate returned by medical staff through the system operation interface. With the goal of minimizing energy consumption, thermal management control parameters are optimized based on the real-time set temperature and real-time set flow rate, and a multi-level heating parameter sequence is output. During the flushing and heating control process using the multi-stage heating parameter sequence, the main outlet of the flushing and heating system is monitored for water flow fluctuations, and congestion risk is predicted based on the monitoring results. If there is a risk of congestion at the main outlet, the redundancy switching mechanism will be activated to switch the water discharge task to the backup outlet.

2. The global intelligent optimization control method for a flushing and heating system as described in claim 1, characterized in that, The method further includes: After receiving a surgical appointment instruction, the flushing and perfusion heating system parses the surgical start timestamp in the surgical appointment instruction to obtain the mode switching node of the surgical support mode. Based on the aforementioned mode switching node, the dynamic self-test time window is traced in reverse to locate the device self-test time node; At the equipment self-inspection time node, perform multi-dimensional component testing on the flushing and heating system and output the equipment self-inspection report; The irrigation and heating system shall be switched to the surgical support mode at the mode switching node only if the self-inspection report of the device concludes that it is qualified in all dimensions.

3. The global intelligent optimization control method for a flushing and heating system as described in claim 1, characterized in that, After verifying the clinical element information, a clinical adaptation safety boundary coupling analysis is performed based on the clinical element information to obtain the safe flushing perfusion temperature threshold and the safe flushing perfusion flow rate threshold. The method includes: Construct a clinical element vector based on the aforementioned clinical element information; After calculating the similarity of multiple elements between the clinical element vector and multiple historical element vectors in the historical flushing control library, P historical element vectors are selected based on a preset similarity threshold. Call upon the P historical flushing and infusion temperatures and P historical flushing and infusion flow rates from the P historical element vectors; Based on the set recurrence frequency of the P historical flushing and infusion temperatures and P historical flushing and infusion flow rates, the P historical flushing and infusion temperatures and P historical flushing and infusion flow rates are weighted and averaged to obtain the initial flushing and infusion temperature threshold and the initial flushing and infusion flow rate threshold. The equipment safety boundary database is retrieved, and the initial flushing injection temperature threshold and the initial flushing injection flow rate threshold are corrected using the constraints of the equipment safety boundary database to obtain the safe flushing injection temperature threshold and the safe flushing injection flow rate threshold.

4. The global intelligent optimization control method for a flushing and heating system as described in claim 3, characterized in that, The clinical information includes the surgical type code, patient body temperature floating point value, irrigation fluid storage temperature value, irrigation fluid pH value, patient age segment, and irrigation fluid viscosity coefficient.

5. The global intelligent optimization control method for a flushing and heating system as described in claim 1, characterized in that, With the goal of minimizing energy consumption, the method optimizes thermal management control parameters based on the real-time set temperature and real-time set flow rate, and outputs a multi-level heating parameter sequence. The method includes: Using the real-time set temperature and real-time set flow rate as joint search conditions, an initial population search is performed to obtain multiple sample heating parameter settings, wherein the sample heating parameter settings include sample power level combination sequence, sample power level time allocation and sample cavity airflow circulation parameters. Energy consumption calculations are performed on the heating parameters of the multiple samples to obtain the energy consumption requirements of the multiple samples. A preset energy consumption baseline is used to traverse the energy consumption requirements of the multiple samples, so as to obtain W benchmark heating parameter settings from the heating parameter settings of the multiple samples; Based on the indicators set by the W benchmark heating parameters, a heating parameter setting space is constructed, and then an optimization envelope space is defined within the heating parameter setting space based on the W benchmark heating parameters. Perform parameter combination iterative search and energy consumption calculation within the optimization envelope space until the convergence condition is reached, and output the multi-level heating parameter sequence.

6. The global intelligent optimization control method for a flushing and heating system as described in claim 1, characterized in that, The method further includes: During the flushing and heating control process using the multi-stage heating parameter sequence, the real-time water temperature and real-time water flow rate of the main outlet are monitored. Based on the real-time outflow velocity and the velocity deviation of the real-time outflow velocity, dynamic PID fine-tuning of the flow pump is performed; Based on the temperature deviation between the real-time outlet water temperature and the real-time set temperature, dynamic PID fine-tuning of the heating power is performed. When the temperature deviation exceeds the preset safety tolerance, the dynamic PID fine-tuning of the flow pump enters the output freeze state. The dynamic PID fine-tuning of heating power and the dynamic PID fine-tuning of flow pump share a deviation threshold triggering rule.

7. The global intelligent optimization control method for a flushing and heating system as described in claim 1, characterized in that, During the flushing and heating control process using the multi-stage heating parameter sequence, the main outlet of the flushing and heating system is monitored for water flow fluctuations, and a congestion risk prediction is performed based on the monitoring results. Prior to this, the method includes: Data backtracking of historical congestion events was performed using congestion risk windows as constraints, including multiple sample time-series flow velocity decline rates, multiple sample time-series pressure rise slopes, multiple sample time-series fluctuation frequencies, and multiple sample event congestion probability values. Congestion risk prediction model based on LSTM network; The congestion risk prediction model is optimized by using the time-series flow velocity decline rate, time-series pressure rise slope, time-series fluctuation frequency, and congestion probability value of the multiple samples as training data.

8. The global intelligent optimization control method for a flushing and heating system as described in claim 7, characterized in that, During the flushing and heating control process using the multi-stage heating parameter sequence, the main outlet of the flushing and heating system is monitored for water flow fluctuations, and congestion risk is predicted based on the monitoring results. The method includes: The congestion risk window monitors the water flow fluctuations at the main outlet using a predefined sliding scale, and obtains multiple real-time flow velocity decrease rates, multiple real-time pressure rise slopes, and multiple real-time fluctuation frequencies. The multiple real-time flow velocity decline rates, multiple real-time pressure rise slopes, and multiple real-time fluctuation frequencies are intermittently loaded into the congestion risk prediction model based on the predefined sliding scale, and congestion risk prediction is performed to output multiple real-time event congestion probabilities. When any of the congestion probabilities of the multiple real-time events meets a preset congestion threshold, a congestion event warning reminder is output. Based on the congestion event early warning and reminder, the redundancy switching mechanism is activated, and the water discharge task is switched to the backup water outlet.

Images