A method and system for intelligent monitoring and compensation of humidification level for a ventilator humidifier

Abstract

The application relates to a humidification level intelligent monitoring and compensation method and system for a ventilator humidifier, comprising the following steps: obtaining operation parameters of a target humidifier and real-time gas flow of a target ventilator; inputting the operation parameters into a preset mass transfer model to obtain a gas humidity value as a monitored humidity value; obtaining historical monitored humidity values and generating a humidification change trend in a target period; obtaining a target humidity value, taking the deviation of the target humidity value and the monitored humidity value as a feedback amount, combining the humidification change trend as a feedforward amount, and generating an adaptive adjustment instruction; in summary, the application obtains real-time operation parameters and gas flow, calculates humidity by using a mass transfer model, combines historical data and a prediction algorithm to generate a change trend, and adjusts heating power based on a composite control strategy, so that the application has the effects of avoiding measurement distortion caused by sensor pollution, improving humidity control precision, and realizing prospective intervention to reduce lag response.

Description

Technical Field

[0001] This application relates to the technical field of humidity monitoring, and in particular to a method and system for intelligent monitoring and compensation of humidification level in a ventilator humidifier. Background Technology

[0002] In the field of respiratory therapy, providing patients with gas at appropriate humidity and temperature is crucial for maintaining normal respiratory mucosal function, preventing sputum accumulation, and related complications. Currently, mainstream humidifiers typically employ a feedback control mechanism based on direct humidity sensors. This involves real-time monitoring of the humidity of the output gas and comparing it to a preset target value to adjust the power of the humidifier's heating element, thus maintaining a constant humidity output. However, this technology, relying on direct physical sensors, has certain drawbacks. For example, the high humidity and bacteria-laden aerosol environment in medical settings can easily lead to sensor probe contamination, drift, or failure, resulting in decreased measurement accuracy or even functional loss. This not only increases equipment maintenance costs and the risk of malfunction but may also cause insufficient or excessive humidification due to distorted monitoring data, potentially damaging the patient's airway or increasing the risk of ventilator-associated pneumonia.

[0003] Furthermore, traditional control strategies are inherently based on a delayed perception and response model, typically initiating adjustment only after detecting a deviation from the target humidity value. This lag makes it difficult to effectively suppress instantaneous fluctuations in humidification levels when faced with disturbances such as dynamic changes in ventilation flow, alterations in patient breathing patterns, or fluctuations in ambient temperature, leading to decreased humidity control accuracy. This is particularly problematic for patients requiring long-term mechanical ventilation, whose individual differences and dynamic changes in their condition often necessitate humidification systems with stronger adaptive capabilities. Current technologies lack predictive analysis of humidity change trends, hindering proactive intervention and failing to meet the demands for precise humidification control. Summary of the Invention

[0004] To address the aforementioned shortcomings, this application provides a method and system for intelligent monitoring and compensation of humidification levels in ventilator humidifiers.

[0005] The above-mentioned objective of this application is achieved through the following technical solution:

[0006] A method for intelligent monitoring and compensation of humidification level in a ventilator humidifier, comprising the following steps:

[0007] The operating parameters of the target humidifier and the real-time gas flow rate of the target ventilator are acquired in real time through the acquisition terminal. The operating parameters include the humidifier tank water temperature and the gas outlet temperature.

[0008] Input the operating parameters into the preset mass transfer model to obtain the gas humidity value, which is then used as the monitoring humidity value;

[0009] Historical humidity values ​​are obtained, and a preset prediction algorithm is used to generate the humidification trend for the target period based on the monitored humidity values, historical humidity values, and real-time gas flow.

[0010] The target humidity value is obtained, and the deviation between the target humidity value and the monitored humidity value is used as the feedback quantity. Combined with the humidification change trend as the feedforward quantity, an adaptive adjustment command is generated through a predefined composite control strategy.

[0011] The heating power of the target humidifier is adjusted according to the generated adaptive adjustment command.

[0012] The second objective of this invention is achieved through the following technical solution:

[0013] A smart monitoring and compensation system for humidification level in a ventilator humidifier includes:

[0014] The data acquisition module is used to acquire the operating parameters of the target humidifier and the real-time gas flow of the target ventilator through the acquisition terminal in real time. The operating parameters include the humidifier tank water temperature and the gas outlet temperature.

[0015] The humidity monitoring module is used to input operating parameters into a preset mass transfer model to obtain gas humidity values, which are then used as monitoring humidity values.

[0016] The trend acquisition module is used to acquire historical humidity values ​​and generate the humidification change trend for the target period based on the monitored humidity values, historical humidity values, and real-time gas flow through a preset prediction algorithm.

[0017] The instruction generation module is used to acquire the target humidity value, use the deviation between the target humidity value and the monitored humidity value as feedback, and combine the humidification change trend as feedforward to generate adaptive adjustment instructions through a predefined composite control strategy.

[0018] The instruction execution module is used to adjust the heating power of the target humidifier according to the generated adaptive adjustment instructions.

[0019] This application also relates to a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described intelligent monitoring and compensation method for humidification level of a ventilator humidifier.

[0020] This application also relates to a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described intelligent monitoring and compensation method for humidification level of a ventilator humidifier.

[0021] In summary, the intelligent monitoring and compensation method and system for humidification level of a ventilator humidifier provided in this application acquires operating parameters and gas flow rate in real time, calculates humidity using a mass transfer model, generates change trends by combining historical data and prediction algorithms, and adjusts heating power based on a composite control strategy. This approach avoids measurement distortion caused by sensor contamination, improves humidity control accuracy, and enables proactive intervention to reduce hysteresis response. Attached Figure Description

[0022] Figure 1 This is a flowchart of an embodiment of an intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to this application;

[0023] Figure 2 This is a flowchart of step S20 in an embodiment of the intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to this application;

[0024] Figure 3 This is a flowchart of step S24 in an embodiment of the intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to this application. Detailed Implementation

[0025] The following is in conjunction with the appendix Figures 1-3 This application will be described in further detail.

[0026] In existing respiratory therapy procedures, humidifier systems rely on direct humidity sensors for feedback control. The technical shortcomings of this system are that the sensors are prone to probe contamination, drift, or failure in high humidity or bacterial aerosol environments, leading to a decrease in measurement accuracy. At the same time, existing control strategies for humidifier systems often employ a hysteresis response mechanism, which cannot promptly correct the humidification level when ventilation flow changes dynamically or when there are environmental disturbances. This results in instantaneous fluctuations in the humidity of the output gas, which in turn affects humidification stability and the safety of the treatment process.

[0027] For example, in the intensive care unit, when a patient is receiving mechanical ventilation due to acute respiratory distress syndrome, the ventilator's ventilation flow rate changes abruptly due to the cough reflex. Aerosols at the humidifier outlet continuously adhere to the surface of the humidity sensor probe, causing the sensor's output signal to gradually deviate from the true value. At this time, because the humidification system fails to recognize the trend of flow rate change, it adjusts the heating power only based on the distorted data, causing the output gas humidity to be significantly lower than the target value in a short period of time. The gas outlet temperature monitoring value also shows abnormal fluctuations, further exacerbating the instability of humidification control.

[0028] If the above problems are not addressed, the continuous fluctuations in humidification levels will lead to dehydration or over-wetting of the respiratory mucosa, impairing the mucociliary clearance function and increasing the risk of sputum accumulation. At the same time, the adjustment bias caused by measurement distortion may keep the humidification system in a non-optimal state for a long time, reduce equipment reliability, and increase the probability of ventilator-associated pneumonia, ultimately affecting the safety of patient treatment and the effectiveness of clinical intervention.

[0029] In one embodiment, such as Figure 1 As shown, this application discloses an intelligent monitoring and compensation method for humidification level in a ventilator humidifier, which specifically includes the following steps:

[0030] S10: Real-time acquisition of the operating parameters of the target humidifier and the real-time gas flow rate of the target ventilator through the acquisition terminal. The operating parameters include the humidifier tank water temperature and the gas outlet temperature.

[0031] In this embodiment, a ventilator humidifier is a medical device used to heat and humidify the gas delivered to the patient by the ventilator to ensure that the gas temperature and humidity meet physiological requirements and protect the patient's respiratory mucosa. A data acquisition terminal refers to a device or module used to acquire the humidifier's operating status and the ventilator's gas flow rate in real time. The data acquisition terminal can be integrated into the humidifier or ventilator, or it can be a separate sensor unit. Operating parameters are physical quantities characterizing the current operating status of the humidifier, including the humidifier tank water temperature and the gas outlet temperature. The humidifier tank water temperature reflects the degree of heating of the water inside the humidifier, while the gas outlet temperature reflects the temperature of the humidified gas when it leaves the humidifier. Real-time gas flow rate refers to the instantaneous volumetric flow rate of the gas delivered by the ventilator to the patient; its dynamic changes directly affect the humidification effect.

[0032] Specifically, the operating parameters of the target humidifier and the real-time gas flow rate of the target ventilator are acquired in real time. These operating parameters include the humidifier tank water temperature and the gas outlet temperature. For example, the humidifier tank water temperature can be obtained by installing a temperature sensor inside the humidifier tank, and the gas outlet temperature can be obtained by installing a temperature sensor at the gas outlet. The real-time gas flow rate can be obtained directly by the flow sensor inside the ventilator, or by installing an independent flow meter on the breathing tubing.

[0033] S20: Input the operating parameters into the preset mass transfer model to obtain the gas humidity value, which is used as the monitoring humidity value;

[0034] In this embodiment, the mass transfer model refers to a mathematical model based on physical principles and empirical formulas, used to describe the process of water vapor being transferred from inside the humidifier to the gas. The gas humidity value is calculated by inputting operating parameters. The monitored humidity value refers to the real-time value obtained by calculating through the mass transfer model to characterize the humidity of the gas at the humidifier outlet.

[0035] Specifically, the acquired operating parameters are input into a preset mass transfer model to obtain the gas humidity value, which is then used as the monitoring humidity value. The mass transfer model can be a corresponding empirical formula, for example, by using a lookup table method or linear interpolation to estimate the gas humidity based on the humidifier water temperature and the gas outlet temperature; or, it can be calculated based on the heat and mass transfer principle using preset fixed coefficients, and the obtained monitoring humidity value reflects the current humidity level of the gas at the humidifier outlet.

[0036] S30: Obtain historical humidity values ​​and generate a humidification trend for the target period based on the monitored humidity values, historical humidity values, and real-time gas flow rate using a preset prediction algorithm.

[0037] In this embodiment, the historical monitoring humidity value refers to the sequence of humidity values ​​calculated and recorded over a period of time to reflect the dynamic evolution of the humidification level; the prediction algorithm refers to the mathematical or statistical method used to analyze historical and current data to infer the trend of humidification level changes over a future period; the humidification change trend refers to the possible changes in the humidity of the humidifier outlet gas during the target period, and further, the humidification change trend includes the direction and magnitude of change.

[0038] Specifically, historical humidity values ​​are obtained, and a preset prediction algorithm is used to generate the humidification trend for the target period based on the monitored humidity values, historical humidity values, and real-time gas flow. Historical humidity values ​​can be obtained by storing a sequence of monitored humidity values ​​over a past period. The prediction algorithm can use statistical methods, such as calculating the average rate of change of humidity over several past time points to predict the direction and magnitude of future humidity changes; or, a regression model based on time series analysis can be used, taking the monitored humidity values, historical humidity values, and real-time gas flow as inputs to predict the short-term humidity trend.

[0039] S40: Obtain the target humidity value, use the deviation between the target humidity value and the monitored humidity value as feedback, and combine the humidification change trend as feedforward, and generate adaptive adjustment commands through a predefined composite control strategy.

[0040] In this embodiment, the target humidity value refers to the ideal humidity level that the humidifier output gas needs to reach, set according to clinical needs or preset standards; the feedback quantity refers to the difference between the monitored humidity value and the target humidity value, used to indicate the degree of deviation between the current humidification level and the desired level; the feedforward quantity refers to the amount by which the control system is adjusted based on the predicted humidification change trend, aiming to preemptively offset possible future humidity deviations; the composite control strategy refers to a control method that combines the advantages of feedback control and feedforward control, aiming to improve the response speed, stability, and anti-interference capability of the control system; the adaptive adjustment command refers to the control signal generated according to the composite control strategy and sent to the humidifier, used to dynamically adjust the heating power of the humidifier.

[0041] Furthermore, the target humidity value can be dynamically adjusted based on the ventilator ventilation mode or patient type information.

[0042] Specifically, the target humidity value is acquired, and the deviation between the target humidity value and the monitored humidity value is used as the feedback quantity. Combined with the humidification change trend as the feedforward quantity, an adaptive adjustment command is generated through a predefined composite control strategy. The target humidity value can be set by medical staff or pre-configured according to the patient's physiological indicators and treatment plan. The feedback quantity can directly calculate the difference between the monitored humidity value and the target humidity value. The feedforward quantity can be quantified according to the predicted humidification change trend. For example, if the humidity is predicted to decrease, the feedforward quantity can be a positive value to increase the heating power in advance. The composite control strategy can be a weighted summation controller that adds the feedback quantity and the feedforward quantity in a fixed ratio to generate an initial adjustment command.

[0043] S50: Adjust the heating power of the target humidifier according to the generated adaptive adjustment command.

[0044] In this embodiment, heating power refers to the output power of the heating element of the humidifier. By adjusting the heating power, the water temperature in the humidification tank can be controlled, thereby affecting the humidification effect.

[0045] Specifically, the heating power of the target humidifier is adjusted according to the generated adaptive adjustment command. For example, the adaptive adjustment command can directly correspond to an increment or decrease in heating power, and the power can be adjusted by controlling the supply voltage or current of the heating element of the humidifier; or, the adaptive adjustment command can be a percentage value used to adjust the output ratio of the current heating power. In this way, the heating power of the humidifier can be dynamically adjusted according to the monitored humidity, historical data and predicted humidity change trends to maintain the humidification level.

[0046] For example, suppose in a medical scenario, a humidifier on a ventilator is providing humidified gas to user A. To ensure the health of user A's respiratory mucosa, the humidity of the humidified gas needs to be maintained at a target humidity value.

[0047] First, the humidifier's operating parameters are acquired in real time through the data acquisition terminal. For example, the humidifier tank water temperature is measured to be 37.5℃ and the gas outlet temperature is measured to be 36.8℃. At the same time, the ventilator's real-time gas flow rate is measured to be 15L / min.

[0048] Next, the operating parameters are input into the preset mass transfer model, which is based on its built-in calculation logic. For example, by considering the saturation characteristics of water vapor at different temperatures and the energy exchange of gas flowing through the humidifier, the current gas humidity value is calculated to be 38 mg / L, which is the monitored humidity value.

[0049] Simultaneously, the system acquires a sequence of monitored humidity values ​​over a past period as historical humidity values, such as the past 10 minutes. Combining the current monitored humidity value of 38 mg / L and the real-time gas flow rate of 15 L / min, the system analyzes these time-series data using a preset prediction algorithm. For example, it finds that the humidity has shown a slight downward trend over the past few minutes, while the gas flow rate is relatively stable. Based on this, the system generates a humidification trend for the target time period. For example, the predicted humidity for the next 30 seconds may continue to decrease at a rate of 0.5 mg / L / min.

[0050] Subsequently, a preset target humidity value is obtained, for example, the target humidity value is set to 40 mg / L. At this time, there is a deviation of -2 mg / L between the monitored humidity value of 38 mg / L and the target humidity value of 40 mg / L, and this deviation is used as a feedback quantity. At the same time, the predicted humidification change trend is quantified as a feedforward quantity. A corresponding adaptive adjustment command is generated by combining the feedback quantity and the feedforward quantity through a predefined composite control strategy. The composite control strategy will comprehensively consider the current deviation and the future trend. For example, if the feedback quantity indicates that the humidity is too low, and the feedforward quantity also predicts that the humidity will decrease further, the composite control strategy will generate an adaptive adjustment command indicating that the heating power needs to be increased.

[0051] Finally, the heating power inside the humidifier is increased according to the adaptive adjustment command. For example, if the command requires an increase of 10% in heating power, the power supply current or voltage of the heating element will be adjusted to increase the heating power output. In this way, the water temperature in the humidification tank will gradually rise, thereby increasing the amount of gas humidification, making the monitored humidity value closer to the target humidity value, and responding in advance to the predicted humidity decline trend.

[0052] Based on the above example, this embodiment obtains the gas humidity value by inputting the operating parameters into a preset mass transfer model, and uses this value as the monitored humidity value, replacing the easily damaged direct humidity sensor. Furthermore, existing control strategies are mostly hysteresis feedback modes, adjusting only after detecting a deviation from the target humidity value, making it difficult to cope with dynamic changes. This embodiment further introduces the prediction of humidification trends: in the above example, by acquiring historical monitored humidity values ​​and combining them with a prediction algorithm, the potential downward trend of humidity in the future can be predicted, thereby generating a feedforward quantity in advance. This feedforward quantity is then combined with the feedback quantity to generate an adaptive adjustment command through a composite control strategy. For example, when a decrease in humidity is predicted, the heating power can be increased in advance, rather than responding only after the humidity actually decreases and a deviation occurs. The introduction of this feedforward control can improve the response speed and control accuracy to dynamic disturbances, effectively avoiding instantaneous fluctuations in humidification levels, thereby achieving precise and stable humidity control.

[0053] In summary, this embodiment overcomes the limitations of existing technologies, such as sensor failure, control lag, and lack of proactive intervention, by employing model-based humidity monitoring, combining historical data and real-time flow for humidification trend prediction, and using a composite control strategy that combines feedback and feedforward. It avoids measurement distortion caused by sensor contamination, improves humidity control accuracy, and achieves proactive intervention to reduce lag response.

[0054] In one embodiment, such as Figure 2 As shown, step S20 includes:

[0055] S21: Input the humidification tank water temperature into the first calculation unit of the mass transfer model, so that the first calculation unit can calculate the maximum saturated absolute humidity value that the gas can reach at the humidification tank water temperature according to its preset saturated water vapor partial pressure calculation formula.

[0056] In this embodiment, the first calculation unit of the mass transfer model is a module within the model, whose function is to calculate the maximum saturated absolute humidity value that the gas can achieve at the humidification tank water temperature. This first calculation unit can be a software module, embedding a formula for calculating the saturated water vapor partial pressure, such as a variant of the Antoine equation or the Clausius-Clapeyron equation. The saturated water vapor partial pressure is obtained through table lookup or direct calculation, and then combined with the total gas pressure to calculate the saturated absolute humidity. Alternatively, it can be a hardware module, such as an application-specific integrated circuit (ASIC) or a programmable gate array (PGA), pre-programmed with the aforementioned calculation logic, capable of rapid response and output of calculation results. The saturated water vapor partial pressure calculation formula describes the partial pressure of water vapor in air at a given temperature when it reaches saturation, and is the basis for calculating the maximum saturated absolute humidity. This formula can employ empirical formulas, such as the Wexler formula or the Goff-Gratch formula, which have high accuracy within a specific temperature range; or it can employ theoretical formulas derived from physical principles, modified with experimental data to adapt to the actual operating conditions inside the humidifier.

[0057] S22: Input the maximum saturated absolute humidity value, real-time gas flow rate and gas outlet temperature into the second calculation unit of the mass transfer model, so that the second calculation unit can calculate the theoretical humidity value when the gas is heated to the gas outlet temperature based on its preset energy and mass conservation law.

[0058] In this embodiment, the second calculation unit of the mass transfer model is a module within the model. Its function is to calculate the theoretical humidity value when the gas is heated to the gas outlet temperature based on the law of conservation of energy and mass. This second calculation unit can be a software module with embedded calculation logic based on enthalpy-humidity diagrams or the thermodynamic properties of moist air, calculating the theoretical humidity value by iteratively or directly solving the energy and mass conservation equations. Alternatively, it can be a software module based on finite element analysis or a simplified computational fluid dynamics model, estimating the theoretical humidity at the gas outlet by simulating the flow and temperature fields inside the humidifier. The law of conservation of energy and mass is a fundamental law in physics, used to describe the transformation and conservation of energy and mass during humidification. In calculations, it can be expressed as a series of simultaneous equations, such as the enthalpy equation for moist air, the water evaporation rate equation, and the gas flow balance equation. Alternatively, the theoretical humidity value can be derived by establishing a control body for the humidifier system and performing balance calculations on the energy and mass flows at the inlet and outlet.

[0059] S23: Input the theoretical humidity value into the third calculation unit of the mass transfer model, so that the third calculation unit reduces the theoretical humidity value by its preset mass transfer efficiency factor to generate the initial humidity calculation value.

[0060] In this embodiment, the third calculation unit of the mass transfer model is a module within the model. Its function is to subtract the theoretical humidity value from the preset mass transfer efficiency factor to generate an initial humidity calculation value. This third calculation unit can be a software module with embedded simple multiplication or functional relationships, multiplying the theoretical humidity value by a mass transfer efficiency factor less than 1. Alternatively, it can be a software module based on lookup tables or empirical curves, dynamically adjusting the mass transfer efficiency factor according to the humidifier's design parameters and operating conditions. The mass transfer efficiency factor is a dimensionless parameter used to characterize the ratio between the actual and ideal mass transfer effects of the humidifier. This factor can be obtained through experimental calibration, such as measuring the ratio of actual humidity to theoretical humidity under different operating conditions; or it can be obtained through theoretical calculations combined with empirical corrections, such as estimation based on the humidifier's geometry and hydrodynamic characteristics.

[0061] S24: Input the initial humidity calculation value and gas outlet temperature into the fourth calculation unit of the mass transfer model, so that the fourth calculation unit uses the gas outlet temperature as a dynamic feedback signal to adjust the initial humidity calculation value and outputs the corrected gas humidity value as the monitoring humidity value.

[0062] In this embodiment, the fourth calculation unit of the mass transfer model is a module within the model. Its function is to use the gas outlet temperature as a dynamic feedback signal to adjust the initial humidity calculation value and output a corrected gas humidity value as the monitored humidity value. The fourth calculation unit can be a software module with an embedded correction algorithm based on proportional-integral-derivative control, or a state estimation algorithm based on Kalman filtering, dynamically adjusting the initial humidity calculation value using real-time changes in the gas outlet temperature. Alternatively, it can be an adaptive correction module based on fuzzy logic or neural networks, adjusting the initial humidity calculation value by learning the relationship between gas outlet temperature and humidity deviation in historical data. The dynamic feedback signal refers to a signal acquired in real-time and used for adjustment. In this embodiment, the gas outlet temperature serves as the dynamic feedback signal, reflecting the actual state of the gas at the humidifier outlet in real time, thereby correcting the humidity value calculated by the model. The gas outlet temperature can be acquired in real-time by a high-precision temperature sensor and transmitted to the fourth calculation unit as a digital signal. Alternatively, multiple temperature sensors can be used to measure at different locations, and the average or weighted average can be taken as a more representative gas outlet temperature to improve the accuracy of the feedback signal.

[0063] Specifically, the humidification tank water temperature is input to the first calculation unit of the mass transfer model. This first calculation unit calculates the maximum saturated absolute humidity value that the gas can achieve at the current humidification tank water temperature based on a preset formula for calculating the partial pressure of saturated water vapor. The maximum saturated absolute humidity value, real-time gas flow rate, and gas outlet temperature are then input to the second calculation unit of the mass transfer model. Based on its preset laws of energy and mass conservation, the second calculation unit comprehensively considers energy exchange and mass transfer within the humidifier to calculate the theoretical humidity value when the gas is heated to the gas outlet temperature. Since actual humidification processes often involve efficiency losses, the theoretical humidity value is further input to the third calculation unit of the mass transfer model. This third calculation unit reduces the theoretical humidity value using its preset mass transfer efficiency factor, thereby generating an initial humidity calculation value that better reflects the actual situation. To further improve the accuracy of humidity monitoring, the initial humidity calculation value and the real-time collected gas outlet temperature are input to the fourth calculation unit of the mass transfer model. The fourth calculation unit uses the gas outlet temperature as a dynamic feedback signal to adjust the initial humidity calculation value in real time, ultimately outputting the corrected gas humidity value as the monitored humidity value.

[0064] To enable those skilled in the art to reproduce this mass transfer model, a specific implementation method is provided below. Specifically, the step of inputting operating parameters into a preset mass transfer model to obtain gas humidity values ​​for monitoring humidity can be achieved in the following way:

[0065] First, the first calculation unit of the mass transfer model can be a software module that stores the calculation formula for the saturated water vapor partial pressure based on the Wexler formula. When the humidification tank water temperature is received, the first calculation unit calculates the saturated partial pressure of water vapor at that temperature according to the formula, and then calculates the maximum saturated absolute humidity value by combining it with atmospheric pressure. Further, the second calculation unit of the mass transfer model can be a software module based on the principle of humid air enthalpy-humidity diagram. After receiving the aforementioned maximum saturated absolute humidity value, real-time gas flow rate, and gas outlet temperature, it iteratively solves the energy and mass conservation equations for humid air to deduce the theoretical humidity value when the gas is heated to 35°C under these conditions. In the first step, the third calculation unit of the mass transfer model can be a multiplier module with a preset mass transfer efficiency factor of 0.95. It multiplies the theoretical humidity value by 0.95 to obtain the initial humidity calculation value. Finally, the fourth calculation unit of the mass transfer model can be a software module based on the Kalman filter algorithm. It receives the initial humidity calculation value and the real-time gas outlet temperature, uses the real-time fluctuation of the gas outlet temperature as dynamic feedback, and combines historical data and model predictions to dynamically adjust the initial humidity calculation value. For example, if the gas outlet temperature shows a slight downward trend, the Kalman filter may consider the initial humidity calculation value to be slightly high and fine-tune it downward, finally outputting the corrected gas humidity value as the monitored humidity value.

[0066] By employing the aforementioned technical solution, the mass transfer process is decomposed into multiple computational units, and dynamic feedback corrections of the laws of conservation of energy and mass, mass transfer efficiency factors, and gas outlet temperature are introduced. This allows the mass transfer model to more accurately simulate the actual physical processes inside the humidifier. Through this multi-unit data processing and correction mechanism, the limitations of a single model under different operating conditions can be effectively compensated, reducing the deviation between the monitored humidity value and the actual humidity. Therefore, the obtained monitored humidity value is more realistic and reliable, providing high-precision input for subsequent intelligent compensation control of humidification levels. This enables more precise adjustment of the humidifier's heating power, avoiding insufficient or excessive humidification due to inaccurate monitoring, ensuring that patients receive appropriate humidified gas, and thus improving patient comfort.

[0067] In one embodiment, such as Figure 3 As shown, step S24 includes:

[0068] S241: Calculate the rate of temperature change based on the gas outlet temperature, and calculate the characteristic parameters that characterize the degree of deviation between the current gas state and the ideal gas state based on the gas outlet temperature, the rate of temperature change, and the real-time gas flow rate.

[0069] In this embodiment, the temperature change rate is calculated based on the gas outlet temperature to obtain the rate at which the gas outlet temperature changes over time. The temperature change rate reflects the dynamic characteristics of the system's thermal state, such as heating or cooling trends. The temperature change rate can be calculated by dividing the temperature difference between consecutive sampling points by the sampling time interval, for example, using the first-order difference method or the moving average difference method; or, using state estimation algorithms such as Kalman filtering, combined with historical temperature data and the system model, to estimate the current temperature change rate, thereby providing a smoother and more accurate estimate even in the presence of measurement noise. Characteristic parameters are calculated to characterize the deviation of the current gas state from the ideal gas state, aiming to quantify the actual gas state and... The difference between theoretical and ideal humidification states; the ideal gas state usually refers to the humidification level at which the gas reaches saturation or near saturation under specific temperature and flow rates; the calculation of characteristic parameters can comprehensively consider gas outlet temperature, temperature change rate, and real-time gas flow rate. For example, a multi-dimensional feature vector can be constructed, which includes the deviation between the gas outlet temperature and the target temperature, the absolute value of the temperature change rate, and the relative deviation between the real-time gas flow rate and the average flow rate; or, by establishing a deviation function based on a physical model, which maps the above input variables to a single and dimensionless deviation index, for example, by calculating the ratio of actual humidification capacity to theoretical maximum humidification capacity, combined with the dynamic influence factors of temperature and flow rate.

[0070] S242: Perform fuzzy logic reasoning on feature parameters through a pre-built fuzzy rule base to generate dynamic correction coefficients, and calculate preliminary correction values ​​based on the dynamic correction coefficients and the initial humidity calculation values.

[0071] In this embodiment, the fuzzy rule base refers to a decision-making system built based on expert knowledge or experience data, used to handle uncertain or fuzzy information. Typically, the fuzzy rule base contains multiple "if-than" rules. The fuzzy logic reasoning process includes fuzzification, fuzzy inference, and defuzzification. Fuzzification converts precise feature parameter values ​​into membership degrees of fuzzy sets; fuzzy inference calculates the activation strength of each rule based on the rule base and membership degrees; and defuzzification converts the fuzzy inference results into precise dynamic correction coefficients. Another implementation method is to use machine learning models such as neural networks or support vector machines to establish a nonlinear mapping relationship between feature parameters and dynamic correction coefficients by learning from a large amount of historical data. The dynamic correction coefficients are... A multiplicative or additive factor is used to adjust the initial humidity calculation value to make it closer to the actual and reasonable humidity level. The generation of the dynamic correction coefficient is a direct output of fuzzy logic reasoning. For example, the dynamic correction coefficient can be a floating-point number between 0 and 1.5. When it deviates from the ideal state, the dynamic correction coefficient will increase or decrease accordingly to compensate for the initial humidity calculation value. The calculation of the preliminary correction value based on the dynamic correction coefficient and the initial humidity calculation value refers to the result after adjusting the initial humidity calculation value according to the dynamic correction coefficient. The calculation method of the preliminary correction value is usually to multiply the initial humidity calculation value by the dynamic correction coefficient, or to add or subtract the correction amount determined by the dynamic correction coefficient to the initial humidity calculation value.

[0072] S243: Perform physiological rationality verification on the preliminary calibration value, and output it as the monitoring humidity value after the verification is passed.

[0073] In this embodiment, the preliminary correction value is physiologically validated to ensure that it is safe, effective, and meets human physiological needs in clinical practice. This validation includes checking whether the humidity value is within a preset safe range and whether the humidity change rate is within an acceptable physiological fluctuation range. Further, the validation can be performed by setting upper and lower thresholds; for example, if the preliminary correction value is lower than a minimum physiological humidity value or higher than a maximum physiological humidity value, a cutoff or alarm is triggered. Alternatively, a corresponding physiological model can be established based on the patient's specific physiological parameters, such as body temperature and respiratory rate, to dynamically evaluate the preliminary correction value and ensure it matches the patient's physiological state.

[0074] Specifically, firstly, the temperature change rate is calculated based on the gas outlet temperature to perceive the dynamic trend of the humidifier outlet gas temperature, rather than just the instantaneous temperature value. Then, combining the gas outlet temperature, temperature change rate, and real-time gas flow rate, a comprehensive characteristic parameter is calculated, which can fully characterize the deviation of the current gas state from the ideal humidification state. By introducing the temperature change rate and real-time gas flow rate, this characteristic parameter can accurately capture the dynamic response of the control system under different operating conditions. For example, when the gas flow rate suddenly increases or decreases, or when the heating power adjustment causes a rapid rise or fall in temperature, this characteristic parameter can promptly reflect the impact of these changes on the humidification effect. Finally, fuzzy logic reasoning is performed on this characteristic parameter using a pre-built fuzzy rule base. The system generates dynamic correction coefficients, enabling dynamic adjustment of correction intensity based on the degree of deviation of the gas state in complex dynamic environments. For example, when the characteristic parameters indicate that the gas state deviates significantly from the ideal value, fuzzy inference generates a larger correction coefficient for greater adjustment; conversely, when the deviation is small, the correction coefficient is smaller. Subsequently, the initial humidity calculation value is adjusted based on the dynamic correction coefficients to obtain a preliminary correction value. The preliminary correction value is then physiologically validated to prevent abnormal humidity values ​​caused by model errors or external interference from being output, thereby protecting the respiratory health of patients. Only humidity values ​​that pass the physiological validation are ultimately output as monitoring humidity values, which improves the reliability and clinical applicability of monitoring results.

[0075] Through the above technical solution, this application introduces a dynamic feedback correction mechanism based on the original mass transfer model's calculation of the initial humidity value. This mechanism comprehensively considers the dynamic changes in gas outlet temperature, the influence of real-time gas flow rate, and the degree of deviation of the system from the ideal state. It also utilizes fuzzy logic for intelligent reasoning to generate adaptive correction coefficients. This correction method enables the monitored humidity value to more accurately reflect the actual gas humidity at the humidifier outlet, especially under dynamic operating conditions, significantly improving the accuracy and robustness of humidification monitoring. Furthermore, the inclusion of physiological rationality verification further enhances the clinical safety and effectiveness of the monitoring results.

[0076] In one embodiment, after step S20, the following is included:

[0077] S201: When the real-time gas flow rate is continuously within the preset stable range and the fluctuation of the gas outlet temperature is less than the preset fluctuation threshold, the current condition is determined to be steady state.

[0078] In this embodiment, step S201 aims to determine that the current operating condition is steady when the real-time gas flow rate and gas outlet temperature are in a stable state. For example, a time window can be set, and the real-time gas flow rate and gas outlet temperature can be continuously monitored within the time window. If all sampled values ​​of the real-time gas flow rate fall within a preset stable range within the entire time window, and the difference between the maximum and minimum values ​​of the gas outlet temperature is less than a preset fluctuation threshold, then the operating condition is determined to be steady. Alternatively, statistical methods can be used, such as calculating the standard deviation or variance of the real-time gas flow rate and gas outlet temperature over a certain period of time.

[0079] S202: Under steady-state conditions, compare the monitored humidity value output by the mass transfer model with the corresponding reference humidity value, calculate the model relative error of the mass transfer efficiency factor, and generate a relative error sequence.

[0080] In this embodiment, step S202 aims to quantify the deviation of the current mass transfer model by comparing it with a known or more accurate reference value when the model is in a steady-state condition. The reference humidity value can be a value directly measured by a high-precision humidity sensor under steady-state conditions. Furthermore, the relative error of the model can be calculated as (monitored humidity value - reference humidity value) / reference humidity value. The relative error sequence is a sequence obtained by storing the relative errors calculated under different steady-state conditions in chronological or operational order. In addition, the reference humidity value can also be the theoretically optimal humidity value calculated based on the humidifier design parameters and ideal mass transfer theory, or a reference value obtained by precise calibration equipment in a laboratory environment. The relative error can be an absolute error or a percentage error, and it is recorded as a relative error sequence.

[0081] S203: Optimize the mass transfer efficiency factor based on the relative error sequence and establish a mapping table between the mass transfer efficiency factor and the operating parameters.

[0082] In this embodiment, step S203 aims to dynamically adjust the mass transfer efficiency factor using accumulated error information, making it more accurately reflect the actual mass transfer process and adaptable to different operating parameters. For example, optimization algorithms such as least squares method and gradient descent method can be used to iteratively adjust the mass transfer efficiency factor with the goal of minimizing the root mean square error of the relative error sequence. At the same time, a mapping relationship table between the optimized mass transfer efficiency factor and the corresponding operating parameters can be established through regression analysis or machine learning methods, such as support vector machine and neural network. Alternatively, a lookup table can be used to obtain the optimized mass transfer efficiency factor through experiments or high-precision measurements under different steady-state conditions and record the corresponding operating parameters. Thus, when the overall system is running, the most suitable mass transfer efficiency factor can be found or interpolated in the mapping relationship table based on the current operating parameters.

[0083] Specifically, the solution in this application calibrates the mass transfer model under the premise of stable system operation by introducing the identification of steady-state operating conditions. Specifically, under steady-state operating conditions, the monitored humidity value output by the mass transfer model is compared with a high-precision reference humidity value to calculate the model relative error of the mass transfer efficiency factor. Furthermore, by continuously collecting relative errors and forming a relative error sequence, deviation information of the mass transfer model under different operating conditions can be obtained. Based on the relative error sequence, the mass transfer efficiency factor is iteratively optimized to more accurately reflect the actual mass transfer process. Furthermore, by establishing a mapping relationship table between the optimized mass transfer efficiency factor and operating parameters, the mass transfer efficiency factor is no longer a static fixed value, but can be dynamically adjusted according to the current operating parameters such as the humidification tank water temperature and gas flow rate.

[0084] Through the above technical solution, this application can effectively solve the problem of decreased accuracy of mass transfer efficiency factor due to environmental changes or equipment aging, thereby improving the accuracy and adaptability of mass transfer model; realize precise monitoring of humidification level of ventilator humidifier, and generate more accurate adaptive adjustment instructions, thereby ensuring the stability and comfort of gas humidity during respiratory therapy, and improving the patient's treatment experience and safety.

[0085] In one embodiment, step S30 includes:

[0086] S31: Time-align the monitored humidity values, historical monitored humidity values, and real-time gas flow rates, and construct a time-series dataset including humidity sequences and corresponding gas flow rates;

[0087] In this embodiment, time alignment of monitored humidity values, historical monitored humidity values, and real-time gas flow rates means ensuring that data points collected from different sources or at different times are consistent or comparable in the time dimension. Specifically, this can be achieved by assigning a precise timestamp to each data point and performing interpolation or resampling as needed. For example, all data can be unified to one sampling point per second. Constructing a time-series dataset including humidity sequences and corresponding gas flow rates means organizing the time-aligned humidity data and gas flow rate data in chronological order to form a structured sequence set. The time-series dataset can be a multidimensional array or table, where each row represents a time point and contains the humidity value and gas flow rate value at that time point, to facilitate subsequent time-series analysis and feature extraction.

[0088] S32: Calculate the first-order difference, moving average, and correlation coefficient with gas flow rate of the monitored humidity value within a set time window based on the time series dataset, and use them as the input feature vector;

[0089] In this embodiment, calculating the first-order difference of the monitored humidity value within a set time window refers to calculating the difference between consecutively monitored humidity values ​​within a predefined time period. For example, the difference between the current humidity value and the previous humidity value can be calculated to reflect the instantaneous rate and direction of humidity change. The set time window can be flexibly set according to actual application needs, such as 5 seconds, 10 seconds, or longer. The moving average refers to averaging the monitored humidity values ​​within a set time window and updating the average value as the time window slides. This helps smooth short-term fluctuations and noise in the data, thereby more accurately revealing the long-term trend or periodic changes in humidity. The correlation coefficient with gas flow rate is a statistical indicator that quantifies the degree of linear or non-linear correlation between the monitored humidity value and the real-time gas flow rate. For example, the Pearson correlation coefficient or Spearman's rank correlation coefficient can be used to assess the correlation between the two to reveal the impact of gas flow rate changes on humidification levels.

[0090] S33: Input the input feature vector into the pre-trained prediction model to obtain the humidification change trend within the target time period. The humidification change trend includes the direction of humidity change and the magnitude of humidity change.

[0091] In this embodiment, the input feature vector refers to a unified numerical vector composed of multiple features such as the first-order difference, moving average, and correlation coefficient calculated above. This input feature vector reflects the dynamic state and trend information of the current humidification environment and serves as the input to the prediction model. The pre-trained prediction model refers to a machine learning model that has been learned and optimized through a large amount of historical data, such as a recurrent neural network, a long short-term memory network, a gated recurrent unit, or an ensemble learning algorithm based on a tree model. This prediction model can identify the complex nonlinear relationship between the input feature vector and the future humidification trend. The humidification trend within the target time period refers to... The prediction model outputs a forecast of how humidification levels will evolve over a future period. This humidification trend can be represented by continuous humidity forecast values ​​or by a direct qualitative description of the change. The humidification trend includes the direction and magnitude of humidity change. The direction of humidity change refers to whether the prediction model determines that the future humidification level will rise, fall, or remain stable, which can usually be determined by analyzing the slope or trend of the forecast value series. The magnitude of humidity change refers to the degree or size of the future humidification level change quantified by the prediction model. For example, it can be expressed as the maximum change in humidity values, the average rate of change, or the relative percentage change within the forecast period.

[0092] Specifically, to ensure data consistency and analyzability, the monitored humidity values ​​from different sources, historical monitored humidity values, and real-time gas flow rates are first time-aligned to ensure that all data points correspond to each other in the time dimension, thereby constructing a complete time-series dataset containing humidity sequences and corresponding gas flow rates. Based on this, to capture the dynamic characteristics of the humidification process, multiple key features are calculated and generated as input feature vectors based on the time-series dataset. The first-order difference of the monitored humidity value within a set time window reflects the instantaneous rate and direction of humidity change, while the moving average helps smooth short-term fluctuations and reveal the long-term trend of humidity. The correlation coefficient with gas flow rate quantifies the degree of mutual influence between humidification level and ventilator gas flow rate. The input feature vectors are then fed into a pre-trained prediction model. This prediction model, by learning complex patterns and correlations in historical data, can predict the humidification trend within a target time period based on the current feature vectors. The humidification trend includes not only the direction of humidity change—whether it is increasing, decreasing, or remaining stable—but also the quantification of the severity of the humidification change, i.e., the magnitude of humidity change.

[0093] Through the above technical solution, this application can overcome the limitations of traditional preset prediction algorithms in handling complex dynamic humidification processes. Specifically, by accurately aligning the monitored humidity value, historical monitored humidity value, and real-time gas flow rate in time, and constructing a structured time-series dataset as the basis for subsequent analysis, the application further captures the dynamic characteristics and intrinsic relationships of the humidification process from multiple dimensions by calculating the first-order difference, moving average, and correlation coefficient with gas flow rate of the monitored humidity value, forming an input feature vector. The input feature vector is then input into a pre-trained prediction model to generate the humidification change trend within the target time period, including a clear direction of humidity change and a quantified magnitude of humidity change. Through this trend prediction process, the effectiveness of feedforward control in the composite control strategy can be improved, enabling the prediction and response to potential fluctuations in humidification levels, thereby improving the real-time performance, stability, and control accuracy of intelligent monitoring and compensation of humidification levels in ventilator humidifiers.

[0094] In one embodiment, the prediction model includes a feature extraction layer, a weight allocation layer, and a regression output layer, and step S33 includes:

[0095] S331: The feature extraction layer extracts the temporal fluctuation features of the humidity sequence and the long-term dependence features between humidity and gas flow rate based on the input feature vector, and fuses them to generate a temporal feature tensor;

[0096] In this embodiment, the feature extraction layer is a component module in the prediction model, used to identify and separate patterns and information meaningful to the prediction task from the original input data. Specifically, its function is to transform the original input feature vector into temporal fluctuation features and long-term dependency features for subsequent processing. For example, a multilayer perceptron can be used to learn the nonlinear combination of input features, or a gated recurrent unit network can be used to capture the temporal dependency in the sequence data. Among them, the temporal fluctuation feature refers to the dynamic change pattern of the humidity sequence within a short time window, reflecting the instantaneous behavior of humidity. For example, the temporal fluctuation feature can be characterized by calculating the local mean square error and autocorrelation coefficient of the humidity sequence, or by using Fourier transform to extract its frequency components. The long-term dependency relationship between humidity and gas flow rate... Long-term dependency features refer to the mutual influence between humidity sequences and gas flow rate sequences over a longer time scale. They help to understand the stable operating modes and potential hysteresis effects of humidification systems under different gas flow rates. For example, long-term dependency features can be characterized by calculating the mutual information value between humidity and gas flow rate, or long-term causal associations between humidity and gas flow rate can be identified using causal inference models. The fusion generation of time-series feature tensors refers to the integration of multiple features extracted from different dimensions or through different methods to form a unified multidimensional data structure, aiming to provide a more comprehensive and richer state representation. For example, the extracted time-series fluctuation features and long-term dependency features can be summed element-wise, or adaptive fusion can be performed through attention networks to generate time-series feature tensors.

[0097] S332: The weight allocation layer calculates and weights the temporal feature tensors based on the attention mechanism to generate a weighted feature tensor;

[0098] In this embodiment, the weight allocation layer is a component module layer in the prediction model, used to evaluate the importance of features at different times or time steps, and assign corresponding weights based on their importance. For example, a cross-attention mechanism based on the Transformer architecture can be used, or a bidirectional recurrent neural network can be used to dynamically adjust feature weights. The attention mechanism is a technique that allows the model to dynamically focus on different parts of the input sequence when processing sequential data. It generates attention weights by calculating the similarity between the query and the key, and then applies the generated attention weights to the values ​​to obtain a weighted representation. For example, multi-head attention can be used to process attention information in different subspaces in parallel, or local attention can be used to limit the attention range. Weighted feature calculation refers to quantifying the contribution of each part of the input features to the current prediction task through attention mechanisms or other methods, and then weighting the features according to these contributions. For example, the attention score can be mapped to between 0 and 1 using the sigmoid function, and then the mapped value is used as a weight to perform element-wise multiplication with the corresponding feature value to achieve weighting. The weighted feature tensor is a feature representation in which the importance of different parts has been explicitly encoded after processing by the weight allocation layer. It not only contains the original feature information, but also highlights the more critical parts for prediction through weights. For example, the weighted feature tensor can be the result of element-wise multiplication of the original temporal feature tensor and importance weights, or the context vector output through a gating mechanism.

[0099] S333: The regression output layer maps the weighted feature tensor to a sequence of humidity prediction values ​​for consecutive time points within the target period, and determines the direction and magnitude of humidity change based on the humidity prediction value sequence.

[0100] In this embodiment, the regression output layer is a component module layer in the prediction model, used to map the processed and weighted feature tensors to the final prediction result, i.e., the humidity prediction value sequence within the target time period. The regression output layer typically consists of one or more fully connected layers, used to perform nonlinear transformations to generate continuous prediction values. For example, the regression output layer can be a fully connected layer containing multiple neurons, each neuron outputting a prediction value at a given time point in the sequence, or a decoder in a sequence-to-sequence model, progressively generating the prediction sequence. The humidity prediction value sequence refers to the continuous prediction of future humidity values ​​within the target time period, providing a detailed trajectory of humidification level changes over time. For example, the humidity prediction value sequence can be a set of humidity values ​​at discrete time points, or a continuous function representation obtained through interpolation. Determining the direction of humidity change refers to judging whether the humidification level is increasing, decreasing, or remaining stable within the target time period. For example, this can be achieved by performing piecewise linear fitting on the humidity prediction value sequence and determining the overall direction based on the majority sign of the slope of each segment; or by comparing the starting and ending points of the sequence. The initial, intermediate, and final values ​​are used to determine the trend. Determining the magnitude of humidity change refers to quantifying the degree of change in humidification level within the target period. For example, the degree of fluctuation can be represented by calculating the variance or standard deviation of the humidity prediction value sequence; or, the difference between the maximum and minimum values ​​in the sequence can be calculated as a relative percentage with the average value of the sequence to quantify the magnitude of change. Among these methods, the slope sign of the sequence is used as a specific way to determine the direction of humidity change. Specifically, by performing linear regression analysis on the humidity prediction value sequence, the slope of the fitted line is obtained. If the slope is positive, it indicates that the humidity is on an upward trend; if it is negative, it indicates that the humidity is on a downward trend; if it is close to zero, it indicates that the humidity is relatively stable. Calculating the relative percentage of the sequence range with the current monitored humidity is another specific way to quantify the magnitude of humidity change. Specifically, the difference between the maximum and minimum values ​​in the humidity prediction value sequence is first calculated, i.e., the sequence range. Then, the sequence range is divided by the monitored humidity value at the current moment and multiplied by 100% to obtain a relative percentage. This relative percentage can intuitively reflect the degree of fluctuation of the humidification level within the target period.

[0101] Specifically, upon receiving the input feature vector, the feature extraction layer first analyzes it to identify and separate the temporal fluctuation characteristics of the humidity sequence and the long-term dependency between humidity and gas flow rate. It then fuses the separated features of different types to generate a temporal feature tensor. Based on this, the weight allocation layer calculates and weights the generated temporal feature tensor using an attention mechanism, thereby dynamically identifying and focusing on the features most influential on the current humidification trend prediction. For example, when gas flow rate changes drastically, the weight allocation layer may assign higher weights to gas flow rate-related features; while when the overall state tends to change... When the humidity is stable, the focus may shift to the temporal fluctuations of humidity itself. Through this attention-weighted approach, the weight allocation layer generates a corresponding weighted feature tensor. Finally, the regression output layer receives the weighted feature tensor and maps it to a sequence of predicted humidity values ​​for consecutive time points within the target period. This sequence of predicted humidity values ​​can reflect the evolution trajectory of future humidification levels. Based on the predicted humidity value sequence, the regression output layer can further determine the direction of humidity change, for example, by analyzing the sign of the slope of the sequence to determine whether it is rising, falling, or stable. At the same time, by calculating the range of the sequence and the relative percentage of the current monitored humidity, the magnitude of humidity change is quantified.

[0102] To enable those skilled in the art to reproduce the prediction model, a specific implementation method is provided below, namely, the prediction model can be implemented using a deep learning architecture. The feature extraction layer can consist of multiple one-dimensional convolutional neural network layers and long short-term memory (LSM) network layers. For example, the one-dimensional convolutional neural network layer can be used to capture local temporal fluctuation features of the humidity sequence, extracting short-term patterns at different scales through convolutional kernels of different sizes; while the LSM network layer can be used to process humidity and gas flow sequences to learn the long-term dependency between humidity and gas flow sequences. Furthermore, the extracted features can be concatenated and fused through a fully connected layer to form a temporal feature tensor; the weight allocation layer can be implemented using a self-attention mechanism. Specifically, the temporal feature tensor is mapped to query, key, and value vectors respectively. The attention weights are obtained by calculating the dot product similarity between the query and the key and normalizing it using the Softmax function; the obtained attention weights are then... The weighted sum of the weight and value vectors is used to generate a weighted feature tensor. The regression output layer can consist of one or more fully connected layers, such as a fully connected layer containing multiple neurons, each neuron outputting a humidity prediction value at a specific time point within the target time period, thus forming a humidity prediction value sequence. When determining the direction of humidity change, the humidity prediction value sequence can be linearly fitted. If the slope of the fitted line is positive, it is judged as an upward trend; if it is negative, it is a downward trend; if it is close to zero, it is a stationary trend. When quantifying the magnitude of humidity change, the difference between the maximum and minimum values ​​in the prediction value sequence can be calculated, i.e., the sequence range. Then, the sequence range is divided by the current monitored humidity value and multiplied by 100% to obtain the relative percentage, which is used as a quantitative indicator of the magnitude of humidity change.

[0103] Through the above technical solutions, the feature extraction layer can capture the temporal fluctuations and long-term dependencies in the humidification process, providing rich input information for prediction; the attention mechanism introduced by the weight allocation layer enables the identification and focusing of key features corresponding to the influence of humidification trends; the regression output layer can generate a continuous sequence of humidity prediction values ​​based on the weighted feature tensor obtained after weight allocation, thereby making the judgment of the direction and magnitude of humidity changes more accurate and reliable; through this hierarchical prediction architecture, a more accurate data foundation can be provided for the precise control of the heating power of the humidifier in the subsequent ventilator, thereby improving the humidification effect and safety.

[0104] In one embodiment, step S40 includes:

[0105] S41: Calculate the real-time deviation between the monitored humidity value and the target humidity value, and use it as a feedback control variable;

[0106] In this embodiment, step S41 aims to obtain the difference between the current humidification level and the desired humidification level. The real-time deviation can be calculated by subtracting the monitored humidity value from the target humidity value, or vice versa, to obtain a signed value. The magnitude of this value reflects the degree of deviation, and the sign reflects the direction of deviation, such as over-humidification or under-humidification. This feedback control quantity refers to the premise used to correct the gap between the current actual output and the desired output. It can be implemented by the subtraction operation module in the digital controller or by the differential amplifier in the analog circuit.

[0107] S42: Extract the direction and magnitude of humidity change within the target time period based on the humidification change trend, and quantify them as feedforward control variables;

[0108] In this embodiment, step S42 aims to intervene in advance using the predicted humidification change trend information to cope with possible future humidity changes. The humidification change trend is usually generated by a prediction model, which includes a direction representing whether the humidity is rising, falling, or remaining stable, as well as the severity of such changes. The humidification change trend is converted into a feedforward control quantity. For example, if the humidity is predicted to drop rapidly, the feedforward control quantity will indicate an increase in heating power in advance. The quantification of the feedforward control quantity can be achieved by encoding the direction of humidity change as positive or negative values ​​and mapping the magnitude of humidity change to the corresponding control increment, for example, by using a lookup table or a preset function.

[0109] S43: Weight allocation of the feedback control quantity and the feedforward control quantity based on the adjustment of real-time gas flow rate;

[0110] In this embodiment, step S43 aims to dynamically adjust the proportions of feedback control and feedforward control in the total control quantity based on the dynamic changes in the external real-time gas flow rate. When the real-time gas flow rate fluctuates significantly, the dynamics are enhanced, and feedforward control is more advantageous due to its predictive nature, enabling it to respond to potential changes more quickly and avoid lag. When the gas flow rate is stable, feedback control can more accurately correct the current deviation. Furthermore, the weight allocation can be achieved through a preset lookup table, a fuzzy logic controller, or an adaptive algorithm.

[0111] S44: Input the weighted feedback control quantity and feedforward control quantity to the PID controller to generate an adaptive adjustment command, and constrain the amplitude and rate of change of the generated adaptive adjustment command.

[0112] In this embodiment, step S44 aims to fuse the feedback and feedforward control quantities after intelligent weight allocation as input to the PID controller to generate an adaptive adjustment command for the heating power. The PID controller is a classic control algorithm that comprehensively considers the current deviation, historical deviation accumulation, and deviation change rate to generate a smooth and responsive control output. Furthermore, amplitude limiting ensures that the adaptive adjustment command does not exceed the physical limit of the humidifier's heating power, preventing overload or invalid output. The rate of change constraint limits the maximum change in the adjustment command per unit time, avoiding drastic fluctuations in heating power, thereby protecting the equipment and preventing discomfort to the patient.

[0113] Specifically, the real-time deviation between the monitored humidity value and the target humidity value is continuously calculated. This real-time deviation serves as the basis for feedback control to correct the difference between the current humidification state and the desired state. Simultaneously, based on the predicted humidification trend, the direction and magnitude of humidity changes within the target time period are extracted and converted into feedforward control variables, aiming to proactively address potential future humidity fluctuations. Furthermore, this embodiment introduces a mechanism that dynamically adjusts the weight allocation between the feedback control and feedforward control variables based on the real-time gas flow rate. When the real-time gas flow rate fluctuates drastically, it is identified as a dynamic operating condition. In this case, the predictive advantage of feedforward control is amplified, and its weight is increased, prioritizing the predicted humidity trend. The system responds quickly enough to effectively avoid humidification lag or insufficiency caused by sudden changes in flow rate. Conversely, when the real-time gas flow rate is in a stable range, the importance of precise correction by feedback control increases, and its weight is increased to ensure the stability and accuracy of the humidification level. Finally, the weighted feedback control quantity and feedforward control quantity are input to the PID controller, which combines current, historical, and future trends to generate corresponding adaptive adjustment commands. To further ensure the safety and stability of the adaptive adjustment commands, amplitude limiting and rate of change constraints are applied to prevent the heating power from exceeding the physical range or causing drastic jumps, thereby ensuring that the humidifier can provide stable humidified gas under various operating conditions.

[0114] Through the above technical solution, this application can significantly improve the control accuracy and response speed of the humidification level of the ventilator humidifier. Especially when the gas flow rate of the ventilator fluctuates frequently or changes rapidly, traditional single feedback or fixed ratio composite control strategies may be difficult to cope with effectively, easily leading to insufficient or excessive humidification. The solution of this embodiment introduces a dynamic weight allocation mechanism based on real-time gas flow to achieve a trade-off between feedback correction of current deviation and feedforward prediction of future trends. Specifically, when the gas flow rate is unstable, the weight of feedforward control is increased, thereby more proactively pre-compensating for potential humidity changes, quickly responding to external disturbances, and avoiding drastic fluctuations in the humidification level. When the gas flow rate is stable, the weight of feedback control is increased to ensure accurate maintenance of the humidification level. Through this adaptive composite control strategy, combined with the adjustment of the PID controller and amplitude limiting and rate of change constraints, not only can the stability of the humidification level be improved and patient discomfort reduced, but the service life of the humidifier components can also be extended, thereby ensuring the effectiveness and safety of the respiratory humidification environment.

[0115] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0116] In one embodiment, an intelligent monitoring and compensation system for humidification level of a ventilator humidifier is provided. This intelligent monitoring and compensation system for humidification level of a ventilator humidifier corresponds one-to-one with the intelligent monitoring and compensation method for humidification level of a ventilator humidifier described in the above embodiment. The intelligent monitoring and compensation system for humidification level of a ventilator humidifier includes:

[0117] The data acquisition module is used to acquire the operating parameters of the target humidifier and the real-time gas flow of the target ventilator through the acquisition terminal in real time. The operating parameters include the humidifier tank water temperature and the gas outlet temperature.

[0118] The humidity monitoring module is used to input operating parameters into a preset mass transfer model to obtain gas humidity values, which are then used as monitoring humidity values.

[0119] The trend acquisition module is used to acquire historical humidity values ​​and generate the humidification change trend for the target period based on the monitored humidity values, historical humidity values, and real-time gas flow through a preset prediction algorithm.

[0120] The instruction generation module is used to acquire the target humidity value, use the deviation between the target humidity value and the monitored humidity value as feedback, and combine the humidification change trend as feedforward to generate adaptive adjustment instructions through a predefined composite control strategy.

[0121] The instruction execution module is used to adjust the heating power of the target humidifier according to the generated adaptive adjustment instructions.

[0122] Specific limitations regarding the intelligent monitoring and compensation system for humidification levels in a ventilator humidifier can be found in the above description of the intelligent monitoring and compensation method for humidification levels in a ventilator humidifier, and will not be repeated here. Each module in the aforementioned intelligent monitoring and compensation system for humidification levels in a ventilator humidifier can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0123] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a method for intelligent monitoring and compensation of humidification level for a ventilator humidifier.

[0124] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements a method for intelligent monitoring and compensation of humidification levels in a ventilator humidifier.

[0125] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for intelligent monitoring and compensation of humidification level in a ventilator humidifier, characterized in that, Including the following steps: The operating parameters of the target humidifier and the real-time gas flow rate of the target ventilator are acquired in real time through the acquisition terminal. The operating parameters include the humidifier tank water temperature and the gas outlet temperature. Input the operating parameters into the preset mass transfer model to obtain the gas humidity value, which is then used as the monitoring humidity value; Historical humidity values ​​are obtained, and a preset prediction algorithm is used to generate the humidification trend for the target period based on the monitored humidity values, historical humidity values, and real-time gas flow. The target humidity value is obtained, and the deviation between the target humidity value and the monitored humidity value is used as the feedback quantity. Combined with the humidification change trend as the feedforward quantity, an adaptive adjustment command is generated through a predefined composite control strategy. The heating power of the target humidifier is adjusted according to the generated adaptive adjustment command.

2. The intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to claim 1, characterized in that: The step of inputting operating parameters into a preset mass transfer model to obtain gas humidity values ​​for monitoring includes the following steps: The humidification tank water temperature is input into the first calculation unit of the mass transfer model, so that the first calculation unit calculates the maximum saturated absolute humidity value that the gas can reach at the humidification tank water temperature according to its preset saturated water vapor partial pressure calculation formula. The maximum saturated absolute humidity value, real-time gas flow rate and gas outlet temperature are input into the second calculation unit of the mass transfer model, so that the second calculation unit can calculate the theoretical humidity value when the gas is heated to the gas outlet temperature based on its preset energy and mass conservation law. The theoretical humidity value is input into the third calculation unit of the mass transfer model, and the third calculation unit reduces the theoretical humidity value by its preset mass transfer efficiency factor to generate the initial humidity calculation value. The initial humidity calculation value and gas outlet temperature are input into the fourth calculation unit of the mass transfer model. The fourth calculation unit uses the gas outlet temperature as a dynamic feedback signal to adjust the initial humidity calculation value and outputs the corrected gas humidity value as the monitoring humidity value.

3. The intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to claim 2, characterized in that: The step of inputting the initial humidity calculation value and the gas outlet temperature into the fourth calculation unit of the mass transfer model, so that the fourth calculation unit uses the gas outlet temperature as a dynamic feedback signal to adjust the initial humidity calculation value and outputs a corrected gas humidity value as the monitored humidity value, includes the following steps: The rate of temperature change is calculated based on the gas outlet temperature, and characteristic parameters that characterize the deviation of the current gas state from the ideal gas state are calculated based on the gas outlet temperature, the rate of temperature change, and the real-time gas flow rate. The feature parameters are subjected to fuzzy logic reasoning through a pre-built fuzzy rule base to generate dynamic correction coefficients, and preliminary correction values ​​are generated based on the dynamic correction coefficients and the initial humidity calculation values. The initial calibration values ​​are physiologically validated, and the results are output as the humidity monitoring values ​​after the validation is passed.

4. The intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to claim 2, characterized in that: After the step of inputting operating parameters into a preset mass transfer model to obtain gas humidity values ​​for monitoring, the following steps are included: When the real-time gas flow rate is continuously within the preset stable range and the fluctuation of the gas outlet temperature is less than the preset fluctuation threshold, the current condition is determined to be steady state. Under steady-state conditions, the monitored humidity value output by the mass transfer model is compared with the corresponding reference humidity value to calculate the model relative error of the mass transfer efficiency factor and generate a relative error sequence. The mass transfer efficiency factor is optimized based on the relative error sequence, and a mapping table between the mass transfer efficiency factor and the operating parameters is established.

5. The intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to claim 1, characterized in that: The step of acquiring historical humidity monitoring values ​​and generating a humidification trend for a target period based on the monitored humidity values, historical humidity monitoring values, and real-time gas flow rate using a preset prediction algorithm includes the following steps: The monitored humidity values, historical monitored humidity values, and real-time gas flow rates are time-aligned, and a time-series dataset including humidity sequences and corresponding gas flow rates is constructed. The first-order difference, moving average, and correlation coefficient with gas flow rate of the monitored humidity value within a set time window are calculated based on the time series dataset and used as the input feature vector. The input feature vector is fed into a pre-trained prediction model to obtain the humidification change trend within the target time period. The humidification change trend includes the direction of humidity change and the magnitude of humidity change.

6. The intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to claim 5, characterized in that: The prediction model includes a feature extraction layer, a weight allocation layer, and a regression output layer. The step of inputting the input feature vector into the pre-trained prediction model to obtain the humidification change trend within the target time period, where the humidification change trend includes the direction and magnitude of humidity change, includes the following steps: The feature extraction layer extracts the temporal fluctuation features of the humidity sequence and the long-term dependence features between humidity and gas flow rate based on the input feature vector, and fuses them to generate a temporal feature tensor; The weight allocation layer calculates and weights the temporal feature tensors based on an attention mechanism to generate a weighted feature tensor. The regression output layer maps the weighted feature tensor to a sequence of humidity prediction values ​​for consecutive time points within the target period, and determines the direction and magnitude of humidity change based on the humidity prediction value sequence.

7. The intelligent monitoring and compensation method for humidification level of a ventilator humidifier according to claim 1, characterized in that: The step of acquiring the target humidity value, using the deviation between the target humidity value and the monitored humidity value as feedback, and combining the humidification change trend as feedforward, to generate an adaptive adjustment command through a predefined composite control strategy includes the following steps: Calculate the real-time deviation between the monitored humidity value and the target humidity value, and use it as a feedback control variable; Based on the humidification trend, the direction and magnitude of humidity change within the target time period are extracted and quantified into feedforward control variables. The weight allocation of the feedback control quantity and the feedforward control quantity is based on the adjustment of the real-time gas flow rate. The weighted feedback control quantity and feedforward control quantity are input to the PID controller to generate an adaptive adjustment command, and the generated adaptive adjustment command is subject to amplitude limiting and rate of change constraint.

8. A smart monitoring and compensation system for humidification level in a ventilator humidifier, characterized in that, include: The data acquisition module is used to acquire the operating parameters of the target humidifier and the real-time gas flow of the target ventilator through the acquisition terminal in real time. The operating parameters include the humidifier tank water temperature and the gas outlet temperature. The humidity monitoring module is used to input operating parameters into a preset mass transfer model to obtain gas humidity values, which are then used as monitoring humidity values. The trend acquisition module is used to acquire historical humidity values ​​and generate the humidification change trend for the target period based on the monitored humidity values, historical humidity values, and real-time gas flow through a preset prediction algorithm. The instruction generation module is used to acquire the target humidity value, use the deviation between the target humidity value and the monitored humidity value as feedback, and combine the humidification change trend as feedforward to generate adaptive adjustment instructions through a predefined composite control strategy. The instruction execution module is used to adjust the heating power of the target humidifier according to the generated adaptive adjustment instructions.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent monitoring and compensation method for humidification level of a ventilator humidifier as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent monitoring and compensation method for humidification level of a ventilator humidifier as described in any one of claims 1-7.

Images