A multivariable fusion control system based on oxygen therapy all-in-one machine

Abstract

The present application relates to the field of oxygen therapy all-in-one machine control system, in particular to a kind of multivariable fusion control system based on oxygen therapy all-in-one machine. Including oxygen therapy all-in-one machine ontology, sensor module, actuator module and control unit. System collects flow, oxygen concentration, outlet temperature, outlet humidity, ambient temperature and blood oxygen saturation and other parameters, under the prerequisite of reserving flow, oxygen concentration, temperature and humidity closed loop, constructs device working condition state quantity, calculates the heat and humidity margin index reflecting the risk of outlet gas heat and humidity fluctuation and the oxygenation index reflecting the risk of oxygenation deficiency, and accordingly the adjustment amount of target flow, target oxygen concentration and target outlet temperature is optimized, target trajectory with changing slope constraint is generated, and is issued to each control loop, realize the collaborative control of multiple physical quantities, reduce the temperature and humidity fluctuation and oxygenation risk, improve the stability and safety of oxygen therapy.

Description

Technical Field

[0001] This invention relates to the field of integrated oxygen therapy machine control systems, specifically a multivariable fusion control system based on an integrated oxygen therapy machine. Background Technology

[0002] Currently, clinically used integrated oxygen therapy machines are relatively mature in terms of functional integration. They can typically perform tasks such as air-oxygen mixing, high-flow delivery, heating and humidification, as well as blood oxygen monitoring and alarms simultaneously. Some devices also incorporate non-invasive ventilation modes and automatic oxygen adjustment functions. From the perspective of individual control loops, the control accuracy of flow rate closed-loop, oxygen concentration closed-loop, and temperature and humidity closed-loop is sufficient for routine treatment needs. However, these improvements are mainly limited to a one-variable-one-loop approach. The interactions between variables are mostly handled through manual experience, simple amplitude limiting, or priority logic, lacking a unified system modeling and coordination mechanism.

[0003] In practical applications, the adjustment response of flow rate and inhaled oxygen concentration is relatively fast, while the heating and humidification process exhibits significant thermal inertia and water / heat storage characteristics. When clinical needs require a rapid increase in flow rate or inhaled oxygen concentration, existing equipment typically adjusts the mixing ratio and fan output immediately. The downstream humidifier often cannot respond synchronously in time, leading to fluctuations in outlet gas temperature and humidity within a short period, and even transitional conditions where the flow rate is high but the temperature has not yet risen and the humidity is significantly low. Existing control structures usually only passively correct for temperature and humidity exceeding limits by triggering alarms or reducing the flow rate, lacking proactive constraints based on heating capacity and current flow rate to predict outlet gas quality before parameter adjustments.

[0004] Similarly, some integrated oxygen therapy machines have introduced automatic oxygen adjustment functions targeting blood oxygen saturation. However, their adjustment logic is mostly based on directly adjusting the target oxygen concentration and / or target flow rate according to the degree to which the current blood oxygen deviates from the target range. While this type of outer-loop adjustment reduces the workload of medical staff to some extent, it usually does not take into account the patient's sensitivity to changes in flow rate, the patient's sensitivity to changes in oxygen concentration, and the blood oxygen trend over a short period of time. When patients have poor lung function reserve or severe ventilation / perfusion imbalance, the same adjustment in flow rate or oxygen concentration may cause large fluctuations in blood oxygen. Existing algorithms often can only compensate afterward when blood oxygen levels drop, which easily leads to fluctuations in blood oxygen levels.

[0005] Regarding multivariate coupling, while existing equipment can simultaneously collect multiple signals such as flow rate, oxygen concentration, outlet temperature, outlet humidity, ambient temperature, and blood oxygen saturation, multivariate fusion is largely limited to simultaneous display on the interface and multi-parameter trend graphs. In terms of control implementation, each closed-loop circuit is essentially tuned independently, and the connections between them are mainly based on empirical logic, such as limiting the maximum flow rate if the temperature is significantly low, or triggering an alarm and locking the flow rate if the oxygen concentration exceeds the range. This rule-based coupling makes it difficult to quantitatively reflect the combined impact of factors such as flow rate variation, heating redundancy, ambient temperature, and pipeline conditions on the outlet gas temperature and humidity, nor can it quantitatively reflect the future risk of the combined effect of flow rate and oxygen concentration on the patient's oxygenation status.

[0006] On the other hand, most existing automatic oxygenation systems determine the direction and magnitude of adjustment based on the average or median blood oxygen saturation within a certain time window, while paying insufficient attention to the dynamic changes in gas pressure, flow rate, and oxygen concentration within a single respiratory cycle or short time scale. When equipment operating conditions change suddenly or treatment parameters are switched rapidly, existing systems often lack a unified index that can simultaneously constrain thermal and humidity fluctuations and oxygenation risks, and can only rely on their own independent alarm limits and protection logic, which can easily lead to instantaneous deterioration of outlet gas quality or brief dips in blood oxygen, while the system struggles to provide a coordinated and consistent adjustment strategy at the algorithm level;

[0007] Furthermore, most existing integrated oxygen therapy machines lack the incorporation of operational status and predictive parameters at the algorithm level. When making adjustment decisions, the control unit primarily relies on current real-time measurements. It lacks explicit quantitative descriptions of questions such as how much the outlet temperature and humidity will deviate under current heating capacity, ambient temperature, and pipeline conditions if the flow rate and oxygen concentration are increased according to the current plan, or whether the blood oxygen level will tend to drop out of the target range within the next few seconds given changes in the patient's current condition and planned parameters. This makes it difficult for existing systems to find a calculable and balancing point between ensuring rapid adjustment and stable output. They can only rely on experience to set conservative slope limits or reserve large safety margins, which affects response speed and is detrimental to energy consumption control and long-term reliability. Summary of the Invention

[0008] The purpose of this invention is to provide a multivariable fusion control system based on an integrated oxygen therapy machine to solve the technical problems mentioned in the background art.

[0009] Based on the above ideas, the present invention provides the following technical solution:

[0010] A multivariable fusion control system based on an integrated oxygen therapy machine includes:

[0011] The main body of the oxygen therapy unit is used to connect with the oxygen source, gas pipeline and patient interface;

[0012] The sensor module is used to collect parameters such as the flow rate and oxygen concentration of the air-oxygen mixing section, the outlet temperature and humidity of the heating and humidification section, as well as the ambient temperature and the patient's blood oxygen saturation.

[0013] The actuator module, including a solenoid valve, a proportional valve, and a heating and humidification unit, is used to regulate flow rate, oxygen concentration, temperature, and humidity.

[0014] The control unit is used to process the parameters collected by the sensor module and issue control commands to the actuator module. The control unit is configured to perform the following steps during system operation:

[0015] S1. After connecting the oxygen therapy unit to the oxygen source, gas pipeline and patient interface and completing sensor calibration, set the target flow rate, target oxygen concentration, target outlet temperature and humidity and target range of blood oxygen saturation through the main control interface, and select either comfort priority mode or treatment priority mode.

[0016] S2. Periodically receive the equipment parameters of flow rate, oxygen concentration, outlet temperature, outlet humidity and ambient temperature collected by the sensor module, construct flow control loop, oxygen concentration control loop and temperature and humidity control loop respectively, and realize basic closed-loop control of flow rate, oxygen concentration and temperature and humidity through the actuator module.

[0017] S3. Continuously receive the patient's blood oxygen saturation signal. When the blood oxygen saturation deviates from the target range, adjust the target oxygen concentration and target flow rate under the safety boundary constraints to keep the blood oxygen saturation within the target range as much as possible.

[0018] S4. Construct an equipment operating condition state vector based on the various equipment parameters. The equipment operating condition state vector includes at least components such as flow rate change ratio, oxygen concentration change ratio, and heating power margin coefficient. Based on the thermal balance relationship between gas mass flow rate and heating power, the control unit calculates the thermal and humidity comfort margin index associated with the predicted deviation of outlet temperature and humidity.

[0019] S5. Based on the deviation between the target lower limit of blood oxygen saturation and the predicted blood oxygen saturation, combined with the flow rate change ratio and oxygen concentration change ratio and the parameters representing the patient's oxygenation sensitivity, the control unit constructs an oxygenation risk index to characterize the risk of insufficient oxygenation or excessive oxygen supply under the adjustment of planned parameters.

[0020] S6. Based on the thermal and humidity comfort margin index and the oxygenation risk index, the control unit establishes an objective function that simultaneously considers thermal and humidity fluctuations, oxygenation risk, and parameter adjustment range. In the comfort-first mode or the treatment-first mode, the control unit performs constrained collaborative optimization on the adjustment amounts of the target flow rate, target oxygen concentration, and target outlet temperature to generate a target trajectory with a change slope constraint. The target trajectory is then sent as a set value to the flow control loop, oxygen concentration control loop, and temperature and humidity control loop to achieve multi-variable fusion control.

[0021] By integrating sensor modules, actuator modules, and multivariable fusion control units within the integrated oxygen therapy machine, the previously independent flow control, oxygen concentration control, and temperature and humidity control are unified into a collaborative framework. On the one hand, while ensuring the stability of basic closed-loop control, a comprehensive assessment of the equipment's operating conditions and the patient's condition is introduced. This ensures that target flow rate, target oxygen concentration, and target outlet temperature and humidity are no longer isolated settings but are dynamically coordinated and generated based on equipment capabilities and patient needs. On the other hand, by selecting comfort-first or treatment-first modes, flexible switching between comfort and emergency needs can be achieved. This makes the equipment suitable for both long-term stable oxygen therapy scenarios and critical care scenarios requiring rapid parameter adjustments, thus improving the system's overall adaptability to complex operating conditions and diverse clinical needs.

[0022] Preferably, the control unit includes a working condition state vector construction and thermal and humidity margin calculation module for executing S4, and further includes:

[0023] Calculate the flow rate change ratio under the current operating conditions, which is the normalized ratio of the difference between the target flow rate and the current flow rate relative to a reference flow rate, used to characterize the intensity of the upcoming flow rate change.

[0024] Calculate the oxygen concentration change ratio under the current operating conditions, which is the normalized ratio of the difference between the target oxygen concentration and the current oxygen concentration relative to a reference oxygen concentration, used to characterize the intensity of the upcoming oxygen supply concentration change.

[0025] Calculate the heating power margin coefficient, which is the ratio of the currently available heating power to the theoretical heating power required at the target flow rate and target outlet temperature based on the gas specific heat, gas mass flow rate and ambient temperature difference. It is used to characterize the redundancy of the heating and humidification unit to the upcoming load changes.

[0026] The flow rate change ratio, oxygen concentration change ratio, and heating power margin coefficient are written as components into the equipment operating state vector. Based on the heat balance relationship and humidification efficiency model, a thermal and humid comfort margin index is obtained to measure the potential deviation of outlet gas temperature and humidity.

[0027] By introducing quantitative parameters such as the intensity of flow rate changes, the intensity of oxygen concentration changes, and heating power margin, and unifying these quantities into the operating condition state vector, the load-bearing capacity of the heating and humidification unit can be predicted before parameter adjustments. Thus, when the flow rate is planned to be increased or the oxygen concentration raised, the system can know in advance whether the heating capacity is sufficient to support the new operating conditions, rather than passively correcting after the outlet temperature and humidity have significantly deviated from the target. This method of constructing the operating condition state vector allows the system to predict and constrain impending temperature and humidity deviations without changing the hardware and software, thereby significantly reducing temperature and humidity fluctuations and condensation problems caused by excessively rapid adjustments, and improving the patient's comfort with inhaled gases.

[0028] Preferably, the control unit calculates the thermal and humidity comfort margin index using the following equation:

[0029] Define the flow rate change ratio a, the oxygen concentration change ratio b, and the heating power margin coefficient c as follows:

[0030]

[0031] Among them, Q target For the target traffic, Q current Q represents the current traffic. ref For reference traffic;

[0032] F O2,target For the target oxygen concentration, F O2,current For the current oxygen concentration, F O2,ref For reference oxygen concentration;

[0033] P h,avail The current available heating power is ρ, where ρ is the gas density and c is the gas density. p For the specific heat capacity at constant pressure, T target For the target outlet temperature, T env Ambient temperature;

[0034] Substituting the flow rate change ratio a, the oxygen concentration change ratio b, and the heating power margin coefficient c into the following formula, the thermal and humidity comfort margin index X is calculated:

[0035]

[0036] γ1 and γ2 are dimensionless weighting coefficients pre-tuned based on equipment calibration and clinical experience, and e is the base of the natural exponential function. The thermal and humidity comfort margin index X increases with the increase of the proportion of flow rate change and the proportion of oxygen concentration change, as well as the decrease of the heating power margin coefficient. It is used to measure the degree of risk of the outlet temperature and humidity deviating from the target range due to the adjustment of instantaneous flow rate and oxygen concentration under a given heating capacity.

[0037] By establishing a thermal and humidity comfort margin index based on the principles of thermal balance and gas mass flow rate, changes in flow rate, oxygen concentration, and residual heating capacity are integrated in a physically coupled manner, rather than through simple linear superposition. This index reflects whether, under current operating conditions, the heating power is sufficient to stabilize the outlet temperature and humidity if the flow rate and oxygen concentration are adjusted as planned. When changes in flow rate and oxygen concentration are large and the heating redundancy is insufficient, this index will increase significantly, thereby driving the upper-level control strategy to automatically reduce the adjustment slope or limit the target value, avoiding dangerous states such as high flow rate and low temperature or high flow rate and low humidity. Through this physically meaningful risk measurement, the actual capacity boundary of the equipment can be more closely approximated, improving the algorithm's ability to identify extreme operating conditions and enhancing safety.

[0038] Preferably, the control unit further includes an oxygenation risk assessment module for performing S5, which includes:

[0039] Based on the current flow rate, oxygen concentration, and patient ventilation status, a pre-set simplified alveolar gas exchange model or an empirical model identified online from equipment operating data is used to predict the patient's predicted blood oxygen saturation within a predetermined time window after the implementation of the planned parameter adjustment.

[0040] The predicted blood oxygen saturation is compared with the lower limit of the target blood oxygen saturation range to obtain the blood oxygen reserve margin, which is used to characterize the risk of hypoxia that may occur without adjusting the parameters.

[0041] By combining the proportion of flow rate change with the proportion of oxygen concentration change, as well as the oxygenation sensitivity coefficient representing the patient's sensitivity to changes in flow rate and oxygen concentration, the blood oxygen reserve margin is corrected, thereby obtaining an oxygenation risk index that comprehensively considers the current oxygenation status, parameter adjustment plan, and individual patient differences.

[0042] The oxygenation risk assessment module links the oxygen supply parameters on the equipment side with the patient's blood oxygenation response using a simplified alveolar gas exchange model or an empirical model, shifting the focus from simply controlling equipment output to addressing patient-centered oxygenation outcomes. This module no longer relies solely on current blood oxygen levels but incorporates planned flow rate and oxygen concentration adjustments to predict blood oxygenation trends over a period of time, thereby proactively identifying potential risks of hypooxygenation or over-oxygenation. This allows for adjustments to the magnitude and rate of oxygenation before significant drops or rises in blood oxygenation occur, helping to reduce the occurrence of severe hypoxia and hyperoxygenation exposures and enhancing the foresight and adaptability of closed-loop control to individual differences.

[0043] Preferably, the oxygenation risk index Y is calculated by the following equation:

[0044] Define the blood oxygen reserve-related component d as:

[0045]

[0046] Among them, S pO2,target,low S represents the lower limit of the target range for blood oxygen saturation. pO2,pred To predict blood oxygen saturation within a predetermined time window;

[0047] Using the proportion of flow rate change (a), the proportion of oxygen concentration change (b), and the patient's oxygenation sensitivity coefficients k1 and k2 to changes in flow rate and oxygen concentration, as well as the nonlinear coupling coefficient μ, the oxygenation risk index Y is calculated according to the following formula:

[0048]

[0049] Wherein, k1 is used to reflect the linear sensitivity of flow rate change to blood oxygen saturation, k2 is used to reflect the linear sensitivity of oxygen concentration change to blood oxygen saturation, μ is used to adjust the nonlinear coupling effect between the proportion of flow rate change and the proportion of oxygen concentration change, and the larger the oxygenation risk index Y value, the higher the residual risk of insufficient oxygenation after considering the adjustment of planned parameters.

[0050] By constructing an oxygenation risk index, the calculation incorporates current blood oxygen reserve, the intensity of planned changes in flow rate and oxygen concentration, and the patient's sensitivity to changes in flow rate and oxygen concentration. A nonlinear coupling term is introduced to reflect the additional impact on oxygenation when flow rate and oxygen concentration change simultaneously. This index can identify potential risks in seemingly minor adjustment combinations; for example, small simultaneous adjustments to flow rate and oxygen concentration may lead to significant blood oxygen fluctuations in certain patients. The quantified oxygenation risk generated in this way allows the control unit to more effectively limit high-risk combinations when planning adjustment trajectories, reducing the probability of blood oxygen overshoot or oscillations.

[0051] Preferably, it also includes a multi-objective collaborative optimization and trajectory generation module for executing S6, and further includes:

[0052] Within each control cycle, the currently calculated thermal and humidity comfort margin index and oxygenation risk index are input into the multi-objective optimization module. Within the preset time prediction step, the increments of the target flow rate, target oxygen concentration, and target outlet temperature are jointly solved.

[0053] In comfort-first mode, the weight of the objective function related to the thermal and humidity comfort margin index is set to be greater than that of the objective function related to the oxygenation risk index, and stricter constraints are imposed on the slope of the change of target flow rate and target oxygen concentration, so as to prioritize the stability of outlet temperature and humidity.

[0054] Under the priority treatment mode, the weight of the objective function related to the oxygenation risk index is set to be greater than the weight of the objective function related to the thermal and humid comfort margin index. Under the premise of meeting the thermal and humid safety boundary, the slope constraints of the target flow rate and target oxygen concentration are relaxed so as to quickly increase or decrease the oxygen supply level.

[0055] Based on the time series of the target flow rate, target oxygen concentration, and target outlet temperature obtained from the solution, a piecewise linear target trajectory is generated, and this target trajectory is used as a set value input into the flow control loop, oxygen concentration control loop, and temperature and humidity control loop.

[0056] Through a multi-objective collaborative optimization and trajectory generation module, the system comprehensively weighs three aspects—thermal and humidity comfort margin, oxygenation risk, and the magnitude of control actions—within each control cycle, configuring different weights and slope constraints for comfort-first and rescue-first modes respectively. On the one hand, in comfort-first mode, the system actively smooths changes in flow rate and oxygen concentration, prioritizing the stability of outlet temperature and humidity, thus suitable for scenarios requiring long-term treatment and high comfort levels. On the other hand, in rescue-first mode, the system allows rapid changes in flow rate and oxygen concentration while ensuring that temperature and humidity remain within safe ranges, to meet the rescue needs in cases of acute oxygenation deterioration. By inputting piecewise linear target trajectories into the basic closed-loop circuit, parameter jumps and frequent overshoots can be effectively avoided, system oscillations can be reduced, and the overall stability and predictability of control can be improved.

[0057] Preferably, the multi-objective optimization module solves for the adjustment amounts of the target flow rate, target oxygen concentration, and target outlet temperature using the following objective function:

[0058]

[0059] Where ΔQ=Q target -Q current , ΔF O2 =F O2,target -F O2,current ΔT=T target -T current ;

[0060] Then, construct the following objective function J:

[0061]

[0062] Where X is the thermal and humidity comfort margin index, Y is the oxygenation risk index, and Q is... ref F O2,ref and T ref The reference scales are flow rate, oxygen concentration, and temperature. w1 to w5 are weighting coefficients used to balance thermal and humidity comfort, oxygenation safety, and control action range.

[0063] By explicitly constructing an objective function that includes thermal and humidity comfort indicators, oxygenation risk indicators, and normalized control increments, patient experience, oxygenation safety, and equipment execution actions are unified within an adjustable and interpretable mathematical framework. On one hand, this allows engineers to quickly adapt to the needs of different departments and patient groups by adjusting different weight parameters, facilitating personalized clinical configuration. On the other hand, by penalizing the square of the control increment, excessively large instantaneous adjustments can be suppressed, constraining the smoothness of control actions at the algorithmic level and avoiding mechanical wear and increased energy consumption caused by frequent large-amplitude movements of the actuator. This objective function form also facilitates the subsequent introduction of more complex constraints or the expansion of new state variables, allowing for upgrades and optimization strategies without altering the overall system architecture, demonstrating good scalability and engineering implementation value.

[0064] The multi-objective optimization module, under the premise of satisfying preset safety constraints and control variable change rate constraints, solves for the control variable increment vector Δu that minimizes the objective function J, and then calculates ΔQ and ΔF. O2 ΔT is used to update the target flow rate, target oxygen concentration, and target outlet temperature within the control unit, thereby completing multivariate fusion optimization control based on the thermal and humid comfort margin index X and the oxygenation risk index Y.

[0065] The technical solution of the present invention may include the following beneficial effects:

[0066] The multivariable fusion control system based on an integrated oxygen therapy machine of this invention, by introducing a state vector, state prediction, and collaborative optimization layer on the basis of existing integrated oxygen therapy machines, elevates the originally independent flow control, oxygen concentration control, and temperature and humidity control from a single-variable mode of adjustment to a multivariable mode that coordinates around the same overall goal. On the one hand, when adjusting the flow rate and oxygen concentration, the heating and humidification capacity and pipeline conditions can be evaluated simultaneously, and the adjustment amplitude and speed can be pre-constrained, so that the temperature, humidity, and pressure fluctuations of the outlet gas are significantly reduced, reducing problems such as sudden temperature changes, airway dryness, and condensation. On the other hand, it can maintain the continuity and stability of the output gas quality even when the equipment operating conditions change abruptly or the treatment parameters are switched rapidly, thereby significantly improving the patient's subjective comfort and the safety margin of the oxygen therapy process.

[0067] By constructing two key indicators—thermal and humidity comfort margin and oxygenation risk—the system organically couples changes in equipment operating conditions with the patient's oxygenation status. It no longer focuses solely on single sensor readings but simultaneously considers whether heating and humidification can withstand the load and whether blood oxygen levels will drop below the safe range in the future. Based on this, the system can predict the risks of temperature and humidity deviations and insufficient oxygenation before parameter adjustments and switch between comfort-first and treatment-first modes according to different clinical needs: the former emphasizes stable and comfortable output gas, suitable for long-term treatment scenarios; the latter, while ensuring temperature and humidity remain within safe ranges, allows for rapid adjustments to flow rate and oxygen concentration, used for acute oxygenation deterioration or resuscitation scenarios. This predictive, collaborative control significantly reduces the probability of blood oxygen fluctuations, parameter oscillations, and frequent manual interventions, improving the system's robustness and self-protection capabilities under complex operating conditions, signal fluctuations, and even partial sensor anomalies.

[0068] The control architecture adopts a layered design. Without altering the existing hardware structure and underlying single-loop control logic of the integrated oxygen therapy machine, it can be implemented simply by adding a prediction and optimization layer in software. Most existing devices can be upgraded to expand functionality, resulting in low engineering implementation costs and high feasibility. The explicit introduction of constraints on the amplitude and slope of control actions in multi-objective optimization helps suppress frequent large-amplitude movements of the actuator, reducing energy consumption and mechanical wear, and extending the lifespan of key components. Simultaneously, the form of the indicators and objective functions reserves space for parameter adjustment and algorithm iteration, allowing for smooth integration of more complex model predictive control or intelligent optimization algorithms to adapt to the individualized needs of different hospitals, departments, and patient groups, demonstrating good scalability and promising prospects for industrial application. Attached Figure Description

[0069] Figure 1 This is a flowchart of a multivariable fusion control system based on an integrated oxygen therapy machine according to the present invention. Detailed Implementation

[0070] Example 1

[0071] like Figure 1 ,

[0072] A multivariable fusion control system based on an integrated oxygen therapy machine includes:

[0073] The main body of the oxygen therapy unit is used to connect with the oxygen source, gas pipeline and patient interface;

[0074] The sensor module is used to collect parameters such as the flow rate and oxygen concentration of the air-oxygen mixing section, the outlet temperature and humidity of the heating and humidification section, as well as the ambient temperature and the patient's blood oxygen saturation.

[0075] The actuator module, including a solenoid valve, a proportional valve, and a heating and humidification unit, is used to regulate flow rate, oxygen concentration, temperature, and humidity.

[0076] The control unit is used to process the parameters collected by the sensor module and issue control commands to the actuator module. The control unit is configured to perform the following steps during system operation:

[0077] S1. After connecting the oxygen therapy unit to the oxygen source, gas pipeline and patient interface and completing sensor calibration, set the target flow rate, target oxygen concentration, target outlet temperature and humidity and target range of blood oxygen saturation through the main control interface, and select either comfort priority mode or treatment priority mode.

[0078] S2. Periodically receive the equipment parameters of flow rate, oxygen concentration, outlet temperature, outlet humidity and ambient temperature collected by the sensor module, construct flow control loop, oxygen concentration control loop and temperature and humidity control loop respectively, and realize basic closed-loop control of flow rate, oxygen concentration and temperature and humidity through the actuator module.

[0079] S3. Continuously receive the patient's blood oxygen saturation signal. When the blood oxygen saturation deviates from the target range, adjust the target oxygen concentration and target flow rate under the safety boundary constraints to keep the blood oxygen saturation within the target range as much as possible.

[0080] S4. Construct an equipment operating condition state vector based on the various equipment parameters. The equipment operating condition state vector includes at least components such as flow rate change ratio, oxygen concentration change ratio, and heating power margin coefficient. Based on the thermal balance relationship between gas mass flow rate and heating power, the control unit calculates the thermal and humidity comfort margin index associated with the predicted deviation of outlet temperature and humidity.

[0081] S5. Based on the deviation between the target lower limit of blood oxygen saturation and the predicted blood oxygen saturation, combined with the flow rate change ratio and oxygen concentration change ratio and the parameters representing the patient's oxygenation sensitivity, the control unit constructs an oxygenation risk index to characterize the risk of insufficient oxygenation or excessive oxygen supply under the adjustment of planned parameters.

[0082] S6. Based on the thermal and humidity comfort margin index and the oxygenation risk index, the control unit establishes an objective function that simultaneously considers thermal and humidity fluctuations, oxygenation risk, and parameter adjustment range. In the comfort-first mode or the treatment-first mode, the control unit performs constrained collaborative optimization on the adjustment amounts of the target flow rate, target oxygen concentration, and target outlet temperature to generate a target trajectory with a change slope constraint. The target trajectory is then sent as a set value to the flow control loop, oxygen concentration control loop, and temperature and humidity control loop to achieve multi-variable fusion control.

[0083] In one specific embodiment, the multivariable fusion control system based on the oxygen therapy integrated machine includes: an air-oxygen mixing module, a high-flow fan module, a heating and humidification module, a patient interface module, a control motherboard, and a power supply module arranged inside the machine casing.

[0084] The air-oxygen mixing module includes: an air inlet, a solenoid air valve, an oxygen inlet, a solenoid oxygen valve, a mixing chamber, and a flow sensor and an oxygen concentration sensor at the mixing chamber outlet. Both the air valve and the oxygen valve are proportional valves, with a drive interface of 0–10V or PWM.

[0085] The high-flow fan module is located downstream of the mixing chamber. It uses a brushless DC fan with a built-in Hall speed sensor. The fan speed signal is input to the control motherboard to assist in estimating the total flow rate.

[0086] The heating and humidification module includes: a water tank with a heating plate, an immersion temperature sensor, an integrated temperature / humidity sensor at the outlet, and a power drive circuit for the heating plate. The heating plate is powered by DC power converted from AC to DC, and PWM power regulation is achieved through MOSFETs or SSRs.

[0087] The patient interface module includes: a flexible heated breathing tubing interface, a nasal cannula or mask interface, and a pressure sensor at the interface for monitoring tubing pressure fluctuations.

[0088] The control motherboard includes: an ARM Cortex-A or Cortex-M processor, an A / D converter, digital input / output ports, and a communication interface for connecting to the user interface board via CAN or RS-485. A real-time operating system runs on the motherboard, dividing control tasks into: data acquisition, basic closed-loop control, operational condition prediction and collaborative optimization, and alarm and recording tasks.

[0089] S1 is implemented by a user interface task and a one-time configuration task. The interface provides target flow rate, target oxygen concentration, target temperature / humidity, target blood oxygen range, and mode selection options. The parameters entered by the user are stored in the global configuration structure.

[0090] S2 is implemented by data acquisition and basic control tasks with a cycle of 10–50ms. It reads data from flow rate, oxygen concentration, temperature and humidity and ambient temperature sensors and sends them to the flow rate PID, oxygen concentration PID and temperature / humidity PID respectively. Each PID directly outputs to the corresponding proportional valve or heating power control.

[0091] In S3, the SpO2 signal from the ear clip or finger clip pulse oximeter is processed, with a sampling period of 200–500 ms. When blood oxygen deviates from the target range, the target oxygen concentration and / or target flow rate are adjusted according to the preset outer loop PI controller. The outer loop only modifies the set value and does not directly output the execution voltage.

[0092] S4–S6 consists of independent operating condition prediction and collaborative optimization tasks running in a period of, for example, 200ms. These tasks read current flow rate, target flow rate, current oxygen concentration, target oxygen concentration, current outlet temperature, ambient temperature, and heating power margin information, calculate the operating condition state vector, further calculate the thermal and humidity comfort margin and oxygenation risk according to the formula, and obtain new target flow rate, target oxygen concentration, and target temperature increments through optimization algorithms. Then, the PID setpoint is refreshed in a slope-limited manner.

[0093] By integrating sensor modules, actuator modules, and multivariable fusion control units within the integrated oxygen therapy machine, the previously independent flow control, oxygen concentration control, and temperature and humidity control are unified into a collaborative framework. On the one hand, while ensuring the stability of basic closed-loop control, a comprehensive assessment of the equipment's operating conditions and the patient's condition is introduced. This ensures that target flow rate, target oxygen concentration, and target outlet temperature and humidity are no longer isolated settings but are dynamically coordinated and generated based on equipment capabilities and patient needs. On the other hand, by selecting comfort-first or treatment-first modes, flexible switching between comfort and emergency needs can be achieved. This makes the equipment suitable for both long-term stable oxygen therapy scenarios and critical care scenarios requiring rapid parameter adjustments, thus improving the system's overall adaptability to complex operating conditions and diverse clinical needs.

[0094] Specifically, the control unit includes a working condition state vector construction and thermal and humidity margin calculation module for executing S4, and also includes:

[0095] Calculate the flow rate change ratio under the current operating conditions, which is the normalized ratio of the difference between the target flow rate and the current flow rate relative to a reference flow rate, used to characterize the intensity of the upcoming flow rate change.

[0096] Calculate the oxygen concentration change ratio under the current operating conditions, which is the normalized ratio of the difference between the target oxygen concentration and the current oxygen concentration relative to a reference oxygen concentration, used to characterize the intensity of the upcoming oxygen supply concentration change.

[0097] Calculate the heating power margin coefficient, which is the ratio of the currently available heating power to the theoretical heating power required at the target flow rate and target outlet temperature based on the gas specific heat, gas mass flow rate and ambient temperature difference. It is used to characterize the redundancy of the heating and humidification unit to the upcoming load changes.

[0098] The flow rate change ratio, oxygen concentration change ratio, and heating power margin coefficient are written as components into the equipment operating state vector. Based on the heat balance relationship and humidification efficiency model, a thermal and humid comfort margin index is obtained to measure the potential deviation of outlet gas temperature and humidity.

[0099] In one embodiment, the control unit is equipped with a working condition state vector construction and thermal and humidity margin calculation module, which reads the following from the basic data acquisition task with a control cycle of 200ms: current total flow rate, target flow rate, current oxygen concentration, target oxygen concentration, current outlet temperature, ambient temperature, and current heating power utilization rate.

[0100] This module first calculates the percentage change in flow rate:

[0101] Select a reference flow rate of 60L / min or the equipment's nominal maximum flow rate, read the current total flow rate and the target flow rate generated by the control outer loop, calculate the difference between the two, and then divide it by the reference flow rate to obtain a normalized value for the flow rate change intensity.

[0102] Next, calculate the percentage change in oxygen concentration:

[0103] Select a reference oxygen concentration of 100% or the upper limit of the equipment, read the current oxygen concentration and the target oxygen concentration given by the outer ring, calculate the difference and divide it by the reference value to obtain the normalized concentration change intensity.

[0104] The heating power margin coefficient is obtained in the following way:

[0105] Read the current heating plate PWM duty cycle or actual output power, and calculate the "current available heating power" by combining it with the maximum power of the heating module;

[0106] Read the current target flow rate and target outlet temperature, and convert the target flow rate into mass flow rate using approximate values ​​of gas density and specific heat capacity;

[0107] Calculate the theoretical power required to reach the target temperature at this flow rate based on the temperature difference between the target outlet temperature and the ambient temperature.

[0108] The ratio of available power to theoretical power is used as the heating power margin coefficient.

[0109] Finally, the aforementioned flow rate change ratio, oxygen concentration change ratio, heating power margin coefficient, and optional pipeline length and ambient temperature are combined into a vector and cached in the control unit for subsequent calculation of thermal and moisture margin indicators. In engineering implementation, each parameter is obtained by looking up tables or using default values ​​from formulas. For example, specific heat capacity can be taken as 1.0 kJ / (kg·K), and density can be corrected with temperature and oxygen concentration, or a simple approximate value can be fixed.

[0110] By introducing quantitative parameters such as the intensity of flow rate changes, the intensity of oxygen concentration changes, and heating power margin, and unifying these quantities into the operating condition state vector, the load-bearing capacity of the heating and humidification unit can be predicted before parameter adjustments. Thus, when the flow rate is planned to be increased or the oxygen concentration raised, the system can know in advance whether the heating capacity is sufficient to support the new operating conditions, rather than passively correcting after the outlet temperature and humidity have significantly deviated from the target. This method of constructing the operating condition state vector allows the system to predict and constrain impending temperature and humidity deviations without changing the hardware and software, thereby significantly reducing temperature and humidity fluctuations and condensation problems caused by excessively rapid adjustments, and improving the patient's comfort with inhaled gases.

[0111] Specifically, the control unit calculates the thermal and humidity comfort margin index using the following equation:

[0112] Define the flow rate change ratio a, the oxygen concentration change ratio b, and the heating power margin coefficient c as follows:

[0113]

[0114] Among them, Q target For the target traffic, Q current Q represents the current traffic. ref For reference traffic;

[0115] F O2,target For the target oxygen concentration, F O2,current For the current oxygen concentration, F O2,ref For reference oxygen concentration;

[0116] P h,avail The current available heating power is ρ, where ρ is the gas density and c is the gas density. p For the specific heat capacity at constant pressure, T target For the target outlet temperature, T env The ambient temperature;

[0117] Substituting the flow rate change ratio a, the oxygen concentration change ratio b, and the heating power margin coefficient c into the following formula, the thermal and humidity comfort margin index X is calculated:

[0118]

[0119] γ1 and γ2 are dimensionless weighting coefficients pre-tuned based on equipment calibration and clinical experience, and e is the base of the natural exponential function. The thermal and humidity comfort margin index X increases with the increase of the proportion of flow rate change and the proportion of oxygen concentration change, as well as the decrease of the heating power margin coefficient. It is used to measure the degree of risk of the outlet temperature and humidity deviating from the target range due to the adjustment of instantaneous flow rate and oxygen concentration under a given heating capacity.

[0120] In one embodiment, the calculation of the thermal and humidity comfort margin index is implemented by a software function. This function receives input parameters such as: target flow rate, current flow rate, reference flow rate, target oxygen concentration, current oxygen concentration, reference oxygen concentration, currently available heating power, target outlet temperature, and ambient temperature, and performs internal calculations.

[0121] The percentage change in oxygen concentration is expressed as a volume fraction, for example, 21% and 60% are represented by 0.21 and 0.60 respectively.

[0122] When the absolute value of the flow rate change ratio increases and the heating margin is low, the value of the first term increases, indicating the risk of forcibly increasing the flow rate when the heating capacity is insufficient;

[0123] When changes in flow rate and oxygen concentration coexist, the second term |ab| characterizes the coupling of these two types of changes, and the exponential factor e −∣c−1∣ This ensures that the coupling effect is most significant when the heating margin is close to 1, and weakens when the heating margin is much greater than or much less than 1. This corresponds to the empirical observation that the system is most sensitive when the heating capacity is just at its critical point. In engineering, γ1 and γ2 are tuned through factory calibration so that X falls within a range that is easy to compare and set thresholds under typical operating conditions, such as 0–10. This function is executed once in each control cycle, and the result is stored in a global variable for use by the upper-level optimization module.

[0124] By establishing a thermal and humidity comfort margin index based on the principles of thermal balance and gas mass flow rate, changes in flow rate, oxygen concentration, and residual heating capacity are integrated in a physically coupled manner, rather than through simple linear superposition. This index reflects whether, under current operating conditions, the heating power is sufficient to stabilize the outlet temperature and humidity if the flow rate and oxygen concentration are adjusted as planned. When changes in flow rate and oxygen concentration are large and the heating redundancy is insufficient, this index will increase significantly, thereby driving the upper-level control strategy to automatically reduce the adjustment slope or limit the target value, avoiding dangerous states such as high flow rate and low temperature or high flow rate and low humidity. Through this physically meaningful risk measurement, the actual capacity boundary of the equipment can be more closely approximated, improving the algorithm's ability to identify extreme operating conditions and enhancing safety.

[0125] Specifically, the control unit further includes an oxygenation risk assessment module for performing S5, which includes:

[0126] Based on the current flow rate, oxygen concentration, and patient ventilation status, a pre-set simplified alveolar gas exchange model or an empirical model identified online from equipment operating data is used to predict the patient's predicted blood oxygen saturation within a predetermined time window after the implementation of the planned parameter adjustment.

[0127] The predicted blood oxygen saturation is compared with the lower limit of the target blood oxygen saturation range to obtain the blood oxygen reserve margin, which is used to characterize the risk of hypoxia that may occur without adjusting the parameters.

[0128] By combining the proportion of flow rate change with the proportion of oxygen concentration change, as well as the oxygenation sensitivity coefficient representing the patient's sensitivity to changes in flow rate and oxygen concentration, the blood oxygen reserve margin is corrected, thereby obtaining an oxygenation risk index that comprehensively considers the current oxygenation status, parameter adjustment plan, and individual patient differences.

[0129] In one embodiment, the oxygenation risk assessment module periodically obtains the following from the data acquisition task: current total flow rate, current oxygen concentration, estimated patient weight or body surface area, current blood oxygen saturation, and target blood oxygen range. The module internally employs a two-layer model:

[0130] The first layer is a simplified alveolar ventilation model: based on the current flow rate and the patient's estimated ventilation needs, the effective flow rate of alveolar gas exchange is estimated, and the alveolar oxygen partial pressure is estimated in combination with the oxygen concentration.

[0131] The second layer is the hemoglobin dissociation curve model: using alveolar oxygen partial pressure as input, the predicted blood oxygen saturation is obtained through approximate formulas or table lookups.

[0132] When planned parameters are adjusted (e.g., target flow rate and target oxygen concentration will change), the module substitutes the "average flow rate and oxygen concentration over a future prediction window (e.g., within 10–20 seconds)" into the above model to obtain the predicted blood oxygen saturation S. pO2,pred The module then compares the predicted oxygen saturation with the lower limit of the target range to obtain the oxygen reserve margin; for example, if the predicted value is 90% and the target lower limit is 92%, the margin is negative, indicating a potential deficiency. To address individual patient differences, a weighted correction is applied to the effects of changes in flow rate and oxygen concentration using a preset or online-adjustable sensitivity coefficient, ensuring a more sensitive oxygen response to the same parameter changes for patients with severe lung function impairment. Those skilled in the art can implement the above prediction module using linear regression or a simple three-dimensional interpolation table based on clinical data.

[0133] The oxygenation risk assessment module links the oxygen supply parameters on the equipment side with the patient's blood oxygenation response using a simplified alveolar gas exchange model or an empirical model, shifting the focus from simply controlling equipment output to addressing patient-centered oxygenation outcomes. This module no longer relies solely on current blood oxygen levels but incorporates planned flow rate and oxygen concentration adjustments to predict blood oxygenation trends over a period of time, thereby proactively identifying potential risks of hypooxygenation or over-oxygenation. This allows for adjustments to the magnitude and rate of oxygenation before significant drops or rises in blood oxygenation occur, helping to reduce the occurrence of severe hypoxia and hyperoxygenation exposures and enhancing the foresight and adaptability of closed-loop control to individual differences.

[0134] Specifically, the oxygenation risk index Y is calculated using the following equation:

[0135] Define the blood oxygen reserve-related component d as:

[0136]

[0137] This value is positive when the predicted blood oxygen level is below the target lower limit, indicating a downward trend; it is negative when it is above the lower limit.

[0138] Among them, S pO2,target,low S represents the lower limit of the target range for blood oxygen saturation. pO2,pred To predict blood oxygen saturation within a predetermined time window;

[0139] Using the proportion of flow rate change (a), the proportion of oxygen concentration change (b), and the patient's oxygenation sensitivity coefficients k1 and k2 to changes in flow rate and oxygen concentration, as well as the nonlinear coupling coefficient μ, the oxygenation risk index Y is calculated according to the following formula:

[0140]

[0141] Oxygenation sensitivity coefficients k1 and k2 are pre-set in the device for different patient types. For example, for patients with COPD, the oxygen concentration sensitivity coefficient can be set higher.

[0142] In the second item, the linear combination of changes in flow rate and oxygen concentration represents the effect of active adjustment on improving blood oxygenation; if the plan is to increase flow rate or oxygen concentration, this item is positive and is subtracted from d, which is equivalent to offsetting part of the risk.

[0143] The third nonlinear term, represented by |ab|, embodies the coupling of the two adjustments. This term is positive when both flow rate and oxygen concentration are adjusted simultaneously, representing the additional uncertainty of the synergistic change. The exponential factor causes the coupling strength to vary with the intensity of change, but it is most pronounced in the intermediate range where the change amplitude is small and the heating margin is moderate. The value of Y is fed into the upper-level optimization module and participates in the objective function calculation along with the heating and moistening margin X. The device can determine k1, k2, μ for each patient type through calibration, or typical values ​​can be fixed empirically.

[0144] Wherein, k1 is used to reflect the linear sensitivity of flow rate change to blood oxygen saturation, k2 is used to reflect the linear sensitivity of oxygen concentration change to blood oxygen saturation, μ is used to adjust the nonlinear coupling effect between the proportion of flow rate change and the proportion of oxygen concentration change, and the larger the oxygenation risk index Y value, the higher the residual risk of insufficient oxygenation after considering the adjustment of planned parameters.

[0145] By constructing an oxygenation risk index, the calculation incorporates current blood oxygen reserve, the intensity of planned changes in flow rate and oxygen concentration, and the patient's sensitivity to changes in flow rate and oxygen concentration. A nonlinear coupling term is introduced to reflect the additional impact on oxygenation when flow rate and oxygen concentration change simultaneously. This index can identify potential risks in seemingly minor adjustment combinations; for example, small simultaneous adjustments to flow rate and oxygen concentration may lead to significant blood oxygen fluctuations in certain patients. The quantified oxygenation risk generated in this way allows the control unit to more effectively limit high-risk combinations when planning adjustment trajectories, reducing the probability of blood oxygen overshoot or oscillations.

[0146] Specifically, it also includes a multi-objective collaborative optimization and trajectory generation module for executing S6, and also includes:

[0147] Within each control cycle, the currently calculated thermal and humidity comfort margin index and oxygenation risk index are input into the multi-objective optimization module. Within the preset time prediction step, the increments of the target flow rate, target oxygen concentration, and target outlet temperature are jointly solved.

[0148] In comfort-first mode, the weight of the objective function related to the thermal and humidity comfort margin index is set to be greater than that of the objective function related to the oxygenation risk index, and stricter constraints are imposed on the slope of the change of target flow rate and target oxygen concentration, so as to prioritize the stability of outlet temperature and humidity.

[0149] Under the priority treatment mode, the weight of the objective function related to the oxygenation risk index is set to be greater than the weight of the objective function related to the thermal and humid comfort margin index. Under the premise of meeting the thermal and humid safety boundary, the slope constraints of the target flow rate and target oxygen concentration are relaxed so as to quickly increase or decrease the oxygen supply level.

[0150] Based on the time series of the target flow rate, target oxygen concentration, and target outlet temperature obtained from the solution, a piecewise linear target trajectory is generated, and this target trajectory is used as a set value input into the flow control loop, oxygen concentration control loop, and temperature and humidity control loop.

[0151] In one embodiment, the multi-objective collaborative optimization and trajectory generation module defines a short prediction step size, such as 2–5 seconds, and discretizes this time window into several sampling points. The module reads from the current state: the current target flow rate, target oxygen concentration, and target temperature, as well as X and Y coordinates.

[0152] The module internally employs simplified single-step optimization instead of complex fully predictive MPC:

[0153] The variables to be solved are set as the increments of the target flow rate, target oxygen concentration, and target temperature within the current control cycle;

[0154] Construct the objective function J (see 7), but in the implementation, it can be approximated by gradient or by enumerating several candidate increments;

[0155] For each candidate increment, simulate the updated trends of X and Y, and calculate J;

[0156] Select the candidate increment with the smallest J as the output for this cycle.

[0157] Mode switching is achieved through a weight table:

[0158] In comfort-first mode, the weights related to heat and humidity are set to larger values ​​in the control parameter table, while the weights related to oxygenation are set to relatively smaller values. At the same time, the upper limits of flow rate and oxygen concentration slope are set to lower values.

[0159] In the treatment-first mode, the weight of oxygenation-related factors is increased and the upper limit of the slope is relaxed, while hard safety boundaries are retained, such as the flow rate not exceeding the rated upper limit and the outlet temperature not exceeding the safe range.

[0160] In trajectory generation, the module updates the setpoint once per cycle and uses linear interpolation to ensure smoothness between setpoint updates. For example, the target flow rate is gradually approximated to the optimized new setpoint in an arithmetic sequence over several cycles to prevent large overshoot in the lower-level PID controller.

[0161] Through a multi-objective collaborative optimization and trajectory generation module, the system comprehensively weighs three aspects—thermal and humidity comfort margin, oxygenation risk, and the magnitude of control actions—within each control cycle, configuring different weights and slope constraints for comfort-first and rescue-first modes respectively. On the one hand, in comfort-first mode, the system actively smooths changes in flow rate and oxygen concentration, prioritizing the stability of outlet temperature and humidity, thus suitable for scenarios requiring long-term treatment and high comfort levels. On the other hand, in rescue-first mode, the system allows rapid changes in flow rate and oxygen concentration while ensuring that temperature and humidity remain within safe ranges, to meet the rescue needs in cases of acute oxygenation deterioration. By inputting piecewise linear target trajectories into the basic closed-loop circuit, parameter jumps and frequent overshoots can be effectively avoided, system oscillations can be reduced, and the overall stability and predictability of control can be improved.

[0162] Specifically, the multi-objective optimization module solves for the adjustment amounts of the target flow rate, target oxygen concentration, and target outlet temperature using the following objective function:

[0163]

[0164] Where ΔQ=Q target -Q current , ΔF O2 =F O2,target -F O2,current ΔT=T target-T current ;

[0165] Where ΔQ, ΔF O2 ΔT corresponds to the total flow rate, oxygen concentration, and outlet temperature that are expected to change in this control cycle. The module learns a reasonable search range from the J comparison results stored in the previous cycle, for example, limiting the flow rate to no more than 5 L / min per cycle, the oxygen concentration to no more than 5%, and the outlet temperature to no more than 1℃.

[0166] Then construct the following objective function J:

[0167]

[0168] The first and second terms take the square of the heat and moisture margin and the risk of oxygenation as the cost. The larger the number, the more the system tends to reduce X and Y by adjusting the control variables.

[0169] The last three terms are represented by the incremental normalized squared value of "action cost" to suppress excessive flow, oxygen concentration and temperature abrupt changes.

[0170] In implementation, the following numerical strategies can be adopted:

[0171] Pre-define a set of candidate increments (e.g., combinations of -1, 0, and +1 unit steps for each variable), iterate through and calculate J for each candidate, and select the smallest; or use the first-order gradient descent method to approximately fine-tune the numerical sensitivity of the increment based on J from the previous cycle.

[0172] Reference value Q ref ,F O2,ref ,T ref The parameters are defined in the factory parameters as 60 L / min, 100%, and 37°C, to unify the increments of different dimensions into dimensionless ones, making the adjustment of w3–w5 more direct. The weights w1–w5 are configured in the software parameter table and can be adjusted in factory calibration or maintenance mode. After obtaining the optimal Δu, the module adds it to the current setpoint to form new target flow rate, target oxygen concentration, and target temperature, and writes them into the global settings for the next cycle's basic closed-loop control task to read and execute.

[0173] Where X is the thermal and humidity comfort margin index, Y is the oxygenation risk index, and Q is the oxygenation risk index. ref F O2,ref and T ref The reference scales are flow rate, oxygen concentration, and temperature. w1 to w5 are weighting coefficients used to balance thermal and humidity comfort, oxygenation safety, and control action range.

[0174] By explicitly constructing an objective function that includes thermal and humidity comfort indicators, oxygenation risk indicators, and normalized control increments, patient experience, oxygenation safety, and equipment execution actions are unified within an adjustable and interpretable mathematical framework. On one hand, this allows engineers to quickly adapt to the needs of different departments and patient groups by adjusting different weight parameters, facilitating personalized clinical configuration. On the other hand, by penalizing the square of the control increment, excessively large instantaneous adjustments can be suppressed, constraining the smoothness of control actions at the algorithmic level and avoiding mechanical wear and increased energy consumption caused by frequent large-amplitude movements of the actuator. This objective function form also facilitates the subsequent introduction of more complex constraints or the expansion of new state variables, allowing for upgrades and optimization strategies without altering the overall system architecture, demonstrating good scalability and engineering implementation value.

[0175] The multi-objective optimization module, under the premise of satisfying preset safety constraints and control variable change rate constraints, solves for the control variable increment vector Δu that minimizes the objective function J, and then calculates ΔQ and ΔF. O2 ΔT is used to update the target flow rate, target oxygen concentration, and target outlet temperature within the control unit, thereby completing multivariate fusion optimization control based on the thermal and humid comfort margin index X and the oxygenation risk index Y.

Claims

1. A multivariable fusion control system based on an integrated oxygen therapy machine, characterized in that, include: The main body of the oxygen therapy unit is used to connect with the oxygen source, gas pipeline and patient interface; The sensor module is used to collect the flow rate and oxygen concentration of the air-oxygen mixing section, the outlet temperature and humidity of the heating and humidification section, as well as the ambient temperature and the patient's blood oxygen saturation. The actuator module, including a solenoid valve, a proportional valve, and a heating and humidification unit, is used to regulate flow rate, oxygen concentration, temperature, and humidity. The control unit is used to process the parameters collected by the sensor module and send control commands to the actuator module; The control unit is configured to perform the following steps during system operation: S1. After connecting the oxygen therapy unit to the oxygen source, gas pipeline and patient interface and completing sensor calibration, set the target flow rate, target oxygen concentration, target outlet temperature and humidity and target range of blood oxygen saturation through the main control interface, and select either comfort priority mode or treatment priority mode. S2. Periodically receive the equipment parameters of flow rate, oxygen concentration, outlet temperature, outlet humidity and ambient temperature collected by the sensor module, construct flow control loop, oxygen concentration control loop and temperature and humidity control loop respectively, and realize basic closed-loop control of flow rate, oxygen concentration and temperature and humidity through the actuator module. S3. Continuously receive the patient's blood oxygen saturation signal. When the blood oxygen saturation deviates from the target range, adjust the target oxygen concentration and target flow rate under the safety boundary constraints to keep the blood oxygen saturation within the target range as much as possible. S4. Construct an equipment operating condition state vector based on the equipment parameters. The equipment operating condition state vector includes at least the flow rate change ratio, the oxygen concentration change ratio, and the heating power margin coefficient. Based on the thermal balance relationship between the gas mass flow rate and the heating power, the control unit calculates the thermal and humidity comfort margin index associated with the predicted deviation of the outlet temperature and humidity. S5. Based on the deviation between the target lower limit of blood oxygen saturation and the predicted blood oxygen saturation, combined with the flow rate change ratio and oxygen concentration change ratio and the parameters representing the patient's oxygenation sensitivity, the control unit constructs an oxygenation risk index to characterize the risk of insufficient oxygenation or excessive oxygen supply under the adjustment of planned parameters. S6. Based on the thermal and humidity comfort margin index and the oxygenation risk index, the control unit establishes an objective function that simultaneously considers thermal and humidity fluctuations, oxygenation risk, and parameter adjustment range. In the comfort-first mode or the treatment-first mode, the control unit performs constrained collaborative optimization on the adjustment amounts of the target flow rate, target oxygen concentration, and target outlet temperature to generate a target trajectory with a change slope constraint. The target trajectory is then sent as a set value to the flow control loop, oxygen concentration control loop, and temperature and humidity control loop to achieve multi-variable fusion control. The control unit calculates the thermal and humidity comfort margin index using the following equation: Define the flow rate change ratio a, the oxygen concentration change ratio b, and the heating power margin coefficient c as follows: Among them, Q target For the target traffic, Q current Q represents the current traffic. ref For reference traffic; F O2,target For the target oxygen concentration, F O2,current For the current oxygen concentration, F O2,ref For reference oxygen concentration; P h,avail The current available heating power is ρ, where ρ is the gas density and c is the gas density. p For the specific heat capacity at constant pressure, T target For the target outlet temperature, T env Ambient temperature; Substituting the flow rate change ratio a, the oxygen concentration change ratio b, and the heating power margin coefficient c into the following formula, the thermal and humidity comfort margin index X is calculated: γ1 and γ2 are dimensionless weighting coefficients pre-tuned based on equipment calibration and clinical experience, and e is the base of the natural exponential function. The thermal and humidity comfort margin index X increases with the increase of the proportion of flow rate change and the proportion of oxygen concentration change, as well as the decrease of the heating power margin coefficient. It is used to measure the degree of risk of the outlet temperature and humidity deviating from the target range due to the adjustment of instantaneous flow rate and oxygen concentration under a given heating capacity.

2. The multivariable fusion control system based on an integrated oxygen therapy machine according to claim 1, characterized in that, The control unit includes a working condition state vector construction and thermal and humidity margin calculation module for executing S4, and also includes: Calculate the flow rate change ratio under the current operating conditions, which is the normalized ratio of the difference between the target flow rate and the current flow rate relative to a reference flow rate, used to characterize the intensity of the upcoming flow rate change. Calculate the oxygen concentration change ratio under the current operating conditions, which is the normalized ratio of the difference between the target oxygen concentration and the current oxygen concentration relative to a reference oxygen concentration, used to characterize the intensity of the upcoming oxygen supply concentration change. Calculate the heating power margin coefficient, which is the ratio of the currently available heating power to the theoretical heating power required at the target flow rate and target outlet temperature based on the gas specific heat, gas mass flow rate and ambient temperature difference. It is used to characterize the redundancy of the heating and humidification unit to the upcoming load changes. The flow rate change ratio, oxygen concentration change ratio, and heating power margin coefficient are written as components into the equipment operating state vector. Based on the heat balance relationship and humidification efficiency model, a thermal and humid comfort margin index is obtained to measure the potential deviation of outlet gas temperature and humidity.

3. The multivariable fusion control system based on an integrated oxygen therapy machine according to claim 2, characterized in that, The control unit also includes an oxygenation risk assessment module for executing S5, which includes: Based on the current flow rate, oxygen concentration, and patient ventilation status, a pre-set simplified alveolar gas exchange model or an empirical model identified online from equipment operating data is used to predict the patient's predicted blood oxygen saturation within a predetermined time window after the implementation of the planned parameter adjustment. The predicted blood oxygen saturation is compared with the lower limit of the target blood oxygen saturation range to obtain the blood oxygen reserve margin, which is used to characterize the risk of hypoxia that may occur without adjusting the parameters. By combining the proportion of flow rate change with the proportion of oxygen concentration change, as well as the oxygenation sensitivity coefficient representing the patient's sensitivity to changes in flow rate and oxygen concentration, the blood oxygen reserve margin is corrected, thereby obtaining an oxygenation risk index that comprehensively considers the current oxygenation status, parameter adjustment plan, and individual patient differences.

4. A multivariable fusion control system based on an integrated oxygen therapy machine according to claim 3, characterized in that, The oxygenation risk index Y is calculated using the following equation: Define the blood oxygen reserve-related component d as: Among them, S pO2,target,low S represents the lower limit of the target range for blood oxygen saturation. pO2,pred To predict blood oxygen saturation within a predetermined time window; Using the proportion of flow rate change (a), the proportion of oxygen concentration change (b), and the patient's oxygenation sensitivity coefficients k1 and k2 to changes in flow rate and oxygen concentration, as well as the nonlinear coupling coefficient μ, the oxygenation risk index Y is calculated according to the following formula: Wherein, k1 is used to reflect the linear sensitivity of flow rate change to blood oxygen saturation, k2 is used to reflect the linear sensitivity of oxygen concentration change to blood oxygen saturation, μ is used to adjust the nonlinear coupling effect between the proportion of flow rate change and the proportion of oxygen concentration change, and the larger the oxygenation risk index Y value, the higher the residual risk of insufficient oxygenation after considering the adjustment of planned parameters.

5. A multivariable fusion control system based on an integrated oxygen therapy machine according to claim 4, characterized in that, It also includes a multi-objective collaborative optimization and trajectory generation module for executing S6, and also includes: Within each control cycle, the currently calculated thermal and humidity comfort margin index and oxygenation risk index are input into the multi-objective optimization module. Within the preset time prediction step, the increments of the target flow rate, target oxygen concentration, and target outlet temperature are jointly solved. In comfort-first mode, the weight of the objective function related to the thermal and humidity comfort margin index is set to be greater than that of the objective function related to the oxygenation risk index, and stricter constraints are imposed on the slope of the change of target flow rate and target oxygen concentration, so as to prioritize the stability of outlet temperature and humidity. Under the priority treatment mode, the weight of the objective function related to the oxygenation risk index is set to be greater than the weight of the objective function related to the thermal and humid comfort margin index. Under the premise of meeting the thermal and humid safety boundary, the slope constraints of the target flow rate and target oxygen concentration are relaxed so as to quickly increase or decrease the oxygen supply level. Based on the time series of the target flow rate, target oxygen concentration, and target outlet temperature obtained from the solution, a piecewise linear target trajectory is generated, and this target trajectory is used as a set value input into the flow control loop, oxygen concentration control loop, and temperature and humidity control loop.

6. A multivariable fusion control system based on an integrated oxygen therapy machine according to claim 5, characterized in that, The multi-objective optimization module solves for the adjustment amounts of the target flow rate, target oxygen concentration, and target outlet temperature using the following objective function: where ΔQ=Q target -Q current , ΔF O2 =F O2,target -F O2,current , ΔT=T target -T current ; Then, construct the following objective function J: Where X is the thermal and humidity comfort margin index, Y is the oxygenation risk index, and Q is... ref F O2,ref and T ref The reference scales are flow rate, oxygen concentration, and temperature. w1 to w5 are weighting coefficients used to balance thermal and humidity comfort, oxygenation safety, and control action range. The multi-objective optimization module, under the premise of satisfying preset safety constraints and control variable change rate constraints, solves for the control variable increment vector Δu that minimizes the objective function J, and then calculates ΔQ and ΔF. O2 ΔT is used to update the target flow rate, target oxygen concentration, and target outlet temperature within the control unit, thereby completing multivariate fusion optimization control based on the thermal and humid comfort margin index X and the oxygenation risk index Y.

Images