Oxygen supply-demand regulation optimization method and system using digital display oxygen flowmeter

Abstract

This invention belongs to the field of oxygen supply regulation, and particularly relates to an oxygen supply and demand regulation optimization method and system using a digital display oxygen flow meter. The method acquires real-time oxygen demand, priority, available resources, oxygen supply quality, and flow meter operation anomaly indicators; initializes a two-way oxygen supply and demand mapping diagram; and generates a supply and demand allocation response strategy by combining two-way supply and demand prediction with an enhanced allocation regulation model. The strategy is implemented on an oxygen supply simulation regulation platform, and indicators are monitored in real time. When indicators do not meet preset constraints, the root cause of the anomaly is located in the mapping diagram based on a causal analysis algorithm. The result is fed back to the enhanced allocation regulation model to dynamically adjust the strategy. Simultaneously, an early warning is issued through a display interface, thereby achieving intelligent optimization scheduling and operation and maintenance of the oxygen supply system, improving resource utilization efficiency and oxygen supply reliability.

Description

Technical Field

[0001] This invention belongs to the field of oxygen supply regulation, and particularly relates to an oxygen supply and demand regulation optimization method and system using a digital display oxygen flow meter. Background Technology

[0002] Currently, centralized medical oxygen supply systems are generally equipped with digital medical oxygen flow meters to monitor terminal oxygen supply, but the value of their data has not been fully explored. Existing technologies mainly rely on regular manual inspections and experience-based judgment, making it difficult to achieve dynamic resource optimization and risk warning. Specifically, the following problems exist: First, the system cannot deeply integrate multi-dimensional indicators such as real-time operating data of the flow meter, patient priority, and multi-concentration oxygen supply capacity, resulting in fragmented scheduling decision-making information; Second, there is a lack of linkage analysis based on supply and demand trend prediction and equipment health status, making it impossible to proactively adjust allocation strategies before resource shortages or equipment performance degradation; Third, when oxygen supply quality fluctuates or flow meter readings are abnormal, existing methods cannot promptly locate the root cause and automatically trigger strategy adjustments and warnings, affecting oxygen supply safety and system stability. Therefore, how to deeply integrate the monitoring and prediction capabilities of digital medical oxygen flow meters to build an integrated system that can achieve intelligent resource scheduling, proactive equipment status maintenance, and closed-loop management of oxygen supply risks is an urgent technical problem to be solved. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention proposes an oxygen supply and demand regulation optimization method and system using a digital display oxygen flow meter. This method acquires real-time oxygen demand, priority, available resources, oxygen supply quality, and flow meter operation anomaly indicators; initializes a two-way oxygen supply and demand mapping diagram; and generates a supply and demand allocation response strategy by combining two-way supply and demand prediction with an enhanced allocation regulation model. The strategy is implemented on an oxygen supply simulation regulation platform, and indicators are monitored in real time. When indicators fail to meet preset constraints, the root cause of the anomaly is located in the mapping diagram based on a causal analysis algorithm. The result is fed back to the enhanced allocation regulation model for dynamic strategy adjustment, and an early warning is issued through a display interface. This achieves intelligent optimization scheduling and operation and maintenance of the oxygen supply system, improving resource utilization efficiency and oxygen supply reliability.

[0004] To achieve the above objectives, the present invention provides the following technical solution: The method for optimizing oxygen supply and demand control using digital oxygen flow meters includes: Acquire real-time oxygen demand indicators, real-time oxygen priority indicators, real-time adjustable resource indicators, real-time oxygen supply quality indicators, and abnormal operation indicators of digital medical oxygen flow meters. Initialize a preset oxygen supply and demand bidirectional mapping diagram, input the real-time oxygen demand index, real-time oxygen priority index, and real-time adjustable resource index into the preset supply and demand bidirectional prediction model and enhanced allocation control model, and combine them with the preset hierarchical mapping to obtain the supply and demand allocation response strategy. On the oxygen supply and demand bidirectional mapping map within the preset oxygen supply simulation and control platform, the system responds to the supply and demand allocation response strategy and monitors real-time oxygen supply quality indicators and abnormal operation indicators of the digital medical oxygen flow meter. When either the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter fails to meet the preset constraint value, the Peter-Clark algorithm is used to analyze and locate the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter on the oxygen supply and demand bidirectional mapping map. The location result is then fed back to the enhanced allocation and control model, and the supply and demand allocation response strategy is adjusted in real time in conjunction with the simulation algorithm. Simultaneously, the location result is fed back to the configured display interface for real-time early warning display.

[0005] Specifically, the real-time oxygen demand indicators include the target label corresponding to each oxygen demand target, the total oxygen demand target corresponding to each oxygen concentration, and the standard oxygen supply flow rate; the real-time oxygen supply priority indicator is constructed from the disease type and severity of each oxygen demand target; the real-time adjustable resource indicator is obtained by measuring the adjustable oxygen quantity under all oxygen supply interfaces corresponding to each oxygen concentration; the real-time oxygen supply quality indicator is constructed by combining the real-time oxygen supply concentration and the real-time oxygen supply flow rate to the supply satisfaction rate; the supply satisfaction rate is constructed by multiplying the ratio of the real-time oxygen supply concentration to the standard oxygen supply concentration, the ratio of the real-time oxygen supply flow rate to the standard oxygen supply flow rate under the corresponding oxygen supply time interval, and the ratio of the actual total oxygen supply to the total oxygen demand target for each oxygen demand target; The abnormal operation index of the digital display medical oxygen flow meter is obtained by combining the operating parameters of each digital display medical oxygen flow meter with the abnormal frequency and lifespan change of the digital display medical oxygen flow meter through an evaluation algorithm with a built-in Bayesian function. It is used to characterize the probability of each digital display medical oxygen flow meter experiencing an abnormality or failure within a preset time period in the future.

[0006] Specifically, the oxygen supply and demand bidirectional mapping diagram includes an oxygen supply diagram, an oxygen demand diagram, supply and demand control nodes, and a medical system interface; The oxygen supply map is constructed using a graph algorithm, combining the connection relationships between each oxygen supply node and oxygen supply equipment. Each oxygen supply node corresponds one-to-one with each oxygen supply interface. The oxygen demand map is constructed from the labels of each target to be supplied with oxygen. The supply and demand control node is connected to each node in the oxygen supply map to obtain the real-time adjustable quantity and total adjustable oxygen quantity of each oxygen supply interface, as well as the predicted real-time adjustable quantity and total adjustable oxygen quantity within a predetermined future time period. The supply and demand control node is connected to the medical system interface to obtain real-time oxygen demand indicators and the disease type and severity of each target to be supplied with oxygen. The supply and demand control node is also connected to each node in the oxygen demand map to allocate the standard oxygen concentration and total standard oxygen flow required for each target to be supplied with oxygen based on its corresponding disease type and severity.

[0007] Specifically, the supply and demand two-way forecasting model includes a supply forecasting sub-model, a demand forecasting sub-model, and a supply and demand discrimination layer; The supply prediction sub-model is built into the connection between the oxygen supply diagram and the supply and demand control node. Based on the real-time adjustable resource index, it predicts the changing trend of the real-time adjustable oxygen quantity of each oxygen supply interface and the total adjustable oxygen quantity of the system within a preset time period. The demand forecasting sub-model is built into the connection between the medical system interface and the supply and demand control node. Based on the real-time oxygen demand index and the real-time oxygen priority index, it predicts the change trend of the total oxygen demand target and the standard oxygen flow rate per unit time of all oxygen-supply targets within a preset time period, and aggregates them into the total oxygen demand change trend.

[0008] Specifically, the supply and demand discrimination layer is connected to the supply forecasting sub-model, the demand forecasting sub-model, and the enhanced allocation and control model, respectively. It is used to compare and detect conflicts between the changing trend of the total adjustable oxygen supply and the changing trend of the total oxygen supply demand in real time. Based on preset discrimination thresholds and matching rules, it identifies potential oxygen supply gaps or resource redundancy risks and their corresponding timestamps within a preset time period, generates a supply-demand imbalance early warning signal, and inputs it into the enhanced allocation and control model. The preset hierarchical mapping is constructed by combining the disease type, the severity of each disease type, the standard oxygen supply concentration under each disease type and corresponding severity, and the hierarchical oxygen supply standard interval with a correlation analysis algorithm. The hierarchical oxygen supply standard interval is constructed by the length of each oxygen supply time interval and the corresponding standard oxygen supply flow rate.

[0009] Specifically, the process of obtaining the supply and demand allocation response strategy includes: The system collects real-time oxygen demand indicators and real-time oxygen priority indicators through the configured medical system interface, and extracts the hierarchical mapping features corresponding to each oxygen supply target. Based on the hierarchical mapping features corresponding to each oxygen supply target, the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate of each oxygen supply target under the current hierarchical mapping are obtained. Collect the trend of oxygen demand changes at different concentrations within a preset historical time period, and combine it with a time series analysis algorithm with seasonal analysis characteristics to construct seasonal oxygen demand change curves at different concentrations and corresponding demand change functions. The seasonal oxygen demand curves and corresponding demand change functions are embedded into the demand prediction sub-model. Combined with real-time collected oxygen demand indicators, the trend of total oxygen demand over a future preset time period is predicted. The total oxygen demand trend includes the total oxygen demand trend of all oxygen targets corresponding to each oxygen concentration, and the trend of the number of oxygen targets under each disease type and corresponding severity.

[0010] Specifically, the process of obtaining the supply and demand allocation response strategy also includes: By collecting data on the total oxygen supply variation over the same time period as oxygen demand and combining it with a time series analysis algorithm with seasonal analysis characteristics, we can construct seasonal curves of total oxygen supply variation at different concentrations and curves of supply variation at each interface. The total supply variation curves of oxygen with different seasonal concentrations and the supply variation curves of each interface are embedded into the supply prediction sub-model to predict the total supply variation trend of each concentration of oxygen and the supply variation trend of each interface within the same preset time period in the future. Based on the predicted trend of total oxygen demand over a future preset time period, the trend of total available oxygen at each concentration over the same future preset time period, and the trend of supply at each interface, the allocation time point when the total target oxygen demand exceeds the real-time available resource index is obtained.

[0011] Specifically, the process of obtaining the supply and demand allocation response strategy also includes: Based on the overall supply trend of each oxygen concentration and the supply trend of each interface after the allocation time point, combined with the overall demand trend of oxygen supply within the same time period, a multi-objective algorithm is used to solve the multi-objective oxygen supply allocation by combining the real-time oxygen supply priority index with the first optimization objective and the first optimization constraint, and the allocation solution is obtained. The first optimization constraint is: the real-time oxygen supply concentration and real-time oxygen supply flow rate corresponding to each oxygen supply target are greater than or equal to the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate corresponding to the current hierarchical mapping. The first optimization objective is: under the condition of satisfying the first optimization constraint, to make the supply satisfaction rate equal to 1 and the oxygen supply risk of all oxygen supply targets minimized. Based on the allocation solution results, when there is an allocation solution result that satisfies the supply satisfaction rate of 1, the standard supply oxygen concentration and the stratified oxygen supply standard range in the stratified mapping corresponding to each node in the oxygen demand graph after the allocation time point are adjusted according to the current solution result. The adjusted stratified mapping is used as the new supply and demand allocation response strategy until the total oxygen demand target is less than or equal to the real-time adjustable resource index, and then the supply and demand allocation response strategy corresponding to the original stratified mapping is restored.

[0012] Specifically, the process of obtaining the supply and demand allocation response strategy also includes: When there is no allocation solution that satisfies the supply satisfaction rate of 1, a forward search is performed using the allocation time point combined with the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate of each target to be supplied under the current hierarchical mapping until an allocation solution that satisfies the supply satisfaction rate of 1 is found, and the last time point of the forward search is taken as the new allocation time point. The hierarchical mapping after the new allocation time point is adjusted using the allocation solution results as a new supply and demand allocation response strategy until the total oxygen demand target is less than or equal to the real-time available resource index, at which point the supply and demand allocation response strategy corresponding to the original hierarchical mapping is restored.

[0013] An oxygen supply and demand control optimization system using a digital display oxygen flow meter includes: The data acquisition module is used to acquire real-time oxygen demand indicators, real-time oxygen priority indicators, real-time adjustable resource indicators, real-time oxygen supply quality indicators, and abnormal operation indicators of the digital medical oxygen flow meter. The enhanced allocation module is used to initialize a preset oxygen supply and demand bidirectional mapping diagram. It combines real-time oxygen demand indicators, real-time oxygen priority indicators, and real-time adjustable resource indicators with a preset supply and demand bidirectional prediction model and an enhanced allocation control model to obtain supply and demand allocation response strategies. The simulation control module is used to respond to the supply and demand allocation response strategy on the oxygen supply and demand bidirectional mapping map within the preset oxygen supply simulation control platform, and to monitor real-time oxygen supply quality indicators and abnormal operation indicators of the digital medical oxygen flow meter. When either the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter fails to meet the preset constraint index value, the Peter-Clark algorithm is used to analyze and locate the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter on the oxygen supply and demand bidirectional mapping map, and the location result is fed back to the enhanced allocation control model. Combined with the simulation algorithm, the supply and demand allocation response strategy is adjusted in real time, and the location result is simultaneously fed back to the configured display interface for real-time early warning display.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention addresses the shortcomings of existing technologies by constructing a two-way oxygen supply and demand mapping diagram. Integrating real-time data sensing, two-way trend prediction, and intelligent reinforcement decision-making, it achieves panoramic digital control and proactive resource optimization of the medical oxygen supply system. Through a two-way supply and demand prediction model and a discrimination layer, it enables forward-looking assessment and early warning of oxygen resource supply and demand trends, significantly improving the foresight and planning of resource allocation and effectively preventing supply gaps or resource waste risks. Secondly, the reinforcement learning-based allocation and control model can dynamically generate and optimize allocation strategies based on real-time monitoring indicators and early warning signals, ensuring a consistently high-priority and stable oxygen supply under complex and ever-changing demand and resource conditions, greatly improving the accuracy of supply and demand matching and overall resource utilization. Finally, this method incorporates real-time oxygen supply quality and equipment malfunctions into closed-loop control, and achieves rapid anomaly location and adaptive strategy correction through causal analysis. This constructs a complete intelligent operation and maintenance closed loop of monitoring-early warning-location-control-verification, greatly enhancing the reliability, safety, and operational response efficiency of the oxygen supply system, and reducing the risks of clinical oxygen use and the workload of manual management. Attached Figure Description

[0015] Figure 1 This is a flowchart of the oxygen supply and demand regulation optimization method using a digital display oxygen flow meter, as described in this invention. Figure 2 This is a three-dimensional model of the digital display medical oxygen flow meter of Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of the forward search in Embodiment 1 of the present invention; Figure 4 This is a block diagram of the oxygen supply and demand regulation optimization system using a digital oxygen flow meter, as described in this invention. Detailed Implementation

[0016] Example 1 Please see Figure 1 and Figure 2 The present invention provides an embodiment of an oxygen supply and demand regulation optimization method using a digital display oxygen flow meter, comprising the following steps: S1. Obtain real-time oxygen demand indicators, real-time oxygen priority indicators, real-time adjustable resource indicators, real-time oxygen supply quality indicators, and abnormal operation indicators of digital medical oxygen flow meter. S2. Initialize a preset oxygen supply and demand bidirectional mapping diagram, input the real-time oxygen demand index, real-time oxygen priority index, and real-time adjustable resource index into the preset supply and demand bidirectional prediction model and enhanced allocation control model, and combine them with the preset hierarchical mapping to obtain the supply and demand allocation response strategy. It should be further explained that the construction and training of the enhanced allocation and control model in this embodiment is achieved through the following steps: First, the state space of the model is defined based on the oxygen supply and demand bidirectional mapping graph. This state space integrates the real-time adjustable resource indicators, real-time oxygen demand indicators, real-time oxygen priority indicators, and the abnormal operation indicators of the digital medical oxygen flow meter. Simultaneously, the action space of the model is defined based on the preset hierarchical mapping and historical control strategy library. The action space is all feasible supply and demand allocation response strategies. Second, the reward function of the model is designed. This function takes maximizing the overall system supply satisfaction rate, minimizing the comprehensive oxygen supply risk, and maximizing resource utilization as its core optimization objectives, and obtains a comprehensive reward value through weighted calculation. Subsequently, based on the deep reinforcement learning framework, the model is trained using a deep Q-network or policy gradient method. During the training process, the model interacts continuously with the environment in the simulation environment generated by the oxygen supply simulation and control platform. Based on the current state, it selects actions and executes the supply and demand allocation response strategy. Then, it receives reward signals and new state feedback from the environment, thereby iteratively updating its neural network parameters, and finally learns the control capability to generate the optimal allocation strategy according to complex dynamic supply and demand states. Once deployed, the model receives early warning signals and real-time monitoring indicators from the supply and demand discrimination layer as state inputs and outputs the optimal supply and demand allocation response strategy in real time. At the same time, the model receives the positioning results from the causal analysis algorithm as negative feedback and dynamically adjusts its strategy network to avoid similar anomalies, thereby achieving an adaptive and continuously optimized intelligent control closed loop.

[0017] S3. On the oxygen supply and demand bidirectional mapping map within the preset oxygen supply simulation and control platform, respond to the supply and demand allocation response strategy and monitor real-time oxygen supply quality indicators and abnormal operation indicators of the digital medical oxygen flow meter. When either the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter fails to meet the preset constraint value, the Peter-Clark algorithm is used to analyze and locate the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter on the oxygen supply and demand bidirectional mapping map, and the location result is fed back to the enhanced allocation and control model. Combined with the simulation algorithm, the supply and demand allocation response strategy is adjusted in real time, and the location result is simultaneously fed back to the configured display interface for real-time early warning display. The simulation algorithm in this embodiment is preferably a discrete event simulation (DES) algorithm. It should be further explained that the real-time oxygen demand index in this embodiment includes the target label corresponding to each oxygen demand target, the total oxygen demand target for each oxygen concentration, and the standard oxygen supply flow rate; the real-time oxygen supply priority index is constructed from the disease type and severity of each oxygen demand target, and the specific implementation process includes: Initial diagnostic and treatment information and real-time monitoring of physiological parameters for each target to be supplied with oxygen are obtained based on medical data records; Based on a pre-defined standardized medical terminology dictionary and natural language processing model, key entity recognition and normalization are performed on the text description in the initial diagnosis and treatment information, and standardized medical feature vectors are extracted and constructed. Based on the standardized medical feature vector, it is input into a pre-trained multi-classification model of diseases. The multi-classification model of diseases is trained based on a deep neural network architecture and using historical labeled medical record data, and outputs the disease type identification result corresponding to each target to be supplied with oxygen. Based on the disease type identification results and the real-time monitored physiological parameters, a preset clinical guideline knowledge base is matched. The clinical guideline knowledge base contains grading rules and thresholds for determining the severity of different disease types. The real-time monitored physiological parameters are quantitatively scored according to the matched grading rules to generate a disease severity assessment result. Based on the symptom type identification results and the symptom severity assessment results, mapping and weighted calculations are performed according to a preset priority quantification comparison table to obtain the final real-time oxygen supply priority index.

[0018] It should be further explained that the real-time adjustable resource index in this embodiment is obtained by measuring the adjustable oxygen quantity under all oxygen supply interfaces corresponding to each oxygen concentration; this embodiment measures the real-time oxygen output and output oxygen flow rate of each interface by configuring a digital display medical oxygen flow meter at the oxygen supply interface. It should be further explained that the real-time oxygen supply quality index in this embodiment is constructed by combining the real-time oxygen supply concentration and the real-time oxygen supply flow rate with the supply satisfaction rate. The supply satisfaction rate is constructed by multiplying the ratio of the real-time oxygen supply concentration to the standard oxygen supply concentration for each target, the ratio of the real-time oxygen supply flow rate to the standard oxygen supply flow rate within the corresponding oxygen supply time interval, and the ratio of the actual total oxygen supply to the total oxygen demand target. It should also be noted that the real-time oxygen supply quality index in this embodiment uses a comprehensive measurement of the real-time oxygen supply concentration and the real-time oxygen supply flow rate combined with the supply satisfaction rate. Its design principle and fundamental motivation are to establish an evaluation system that can comprehensively, accurately, and dynamically reflect the actual effect of medical oxygen supply. Oxygen concentration is the core parameter determining the physiological effect of oxygen therapy and is directly related to the treatment effect and patient safety. The oxygen supply flow rate determines the total amount of oxygen delivered to the patient's respiratory system per unit time and is key to meeting their ventilation needs per minute. Monitoring either parameter alone has significant limitations; for example, accurate concentration but insufficient flow rate will lead to insufficient total oxygen supply and cannot effectively correct hypoxia. Therefore, this method innovatively introduces the supply satisfaction rate as a composite core indicator for overall correction. This indicator assesses concentration accuracy by calculating the ratio of real-time concentration to standard concentration, assesses flow rate stability by calculating the ratio of real-time flow rate to standard flow rate, and assesses total supply completion by calculating the ratio of actual total supply to total demand. Finally, the product of these three key ratios is used as the comprehensive evaluation criterion. The core motivation for this multi-dimensional integrated measurement is to build a monitoring mechanism that can sensitively capture any defects in oxygen supply quality caused by equipment performance fluctuations, pipeline abnormalities, or scheduling errors. Once this indicator declines, the system can immediately identify which dimension—concentration, flow rate, or total supply—has deviated, thereby driving the control system to perform precise source tracing and intervention. This ensures that oxygen delivery closely matches each patient's individualized and potentially dynamically changing medical needs throughout the entire process, ultimately achieving a fundamental improvement from simple oxygen supply to high-quality, precise, and safe therapeutic delivery.

[0019] It should be further explained that the abnormal operation index of the digital display medical oxygen flow meter in this embodiment is obtained by combining the operating parameters of each digital display medical oxygen flow meter with the abnormal frequency and lifespan change of the digital display medical oxygen flow meter, through an evaluation algorithm with a built-in Bayesian function. It is used to characterize the probability of each digital display medical oxygen flow meter experiencing an abnormality or failure within a preset time period in the future. The specific construction and implementation process includes: S101. Based on the real-time data interface between the hospital equipment operation and maintenance database and the digital display medical oxygen flow meter, acquire the historical operating parameter time-series data, historical abnormal event records, and cumulative working time data of each target flow meter; the historical operating parameters include at least the historical oxygen output flow rate, the deviation between the historical displayed value and the set value, and the historical sensor signal stability data. S102. Based on a preset statistical analysis time window, feature extraction is performed on the historical operating parameter time series data to obtain an operating feature vector including mean, variance, and trend slope. S103. Based on the historical abnormal event records, count the number of abnormal occurrences of each target flow table within the statistical analysis time window to obtain the historical abnormal frequency. S104. Based on the cumulative working time data and the preset equipment life benchmark model, calculate the offset of the current life stage relative to the standard life and obtain the life change. S105. Based on the Bayesian network structure learning algorithm, construct an initial fault prediction network topology; the input of the network topology is the operating feature vector, historical abnormal frequency and lifespan change of the digital display medical oxygen flow meter corresponding to each node in the oxygen demand graph, and the output is the predicted probability of the corresponding digital display medical oxygen flow meter experiencing an abnormality or fault within a preset time period in the future. S106. The initial fault prediction network is trained with parameters based on the historical labeled dataset. The historical labeled dataset is constructed from the time series data of the historical operating parameters, the corresponding historical anomaly frequency, the lifetime change amount, and the labels of the actual anomalies or faults. The maximum likelihood estimation method is used to iteratively update the conditional probability table parameters of each node in the Bayesian network through the expectation-maximization algorithm until the model converges, and a fully trained Bayesian evaluation model is obtained. S107. During the real-time operation phase, based on the fully trained Bayesian evaluation model, probabilistic reasoning is performed on the real-time operating parameters input to each digital display medical oxygen flow meter, the real-time calculated abnormal frequency and lifespan change, and the probability value representing the occurrence of abnormality or failure within a preset time period is obtained and output in real time as the final abnormal operation index of the digital display medical oxygen flow meter.

[0020] It should be further explained that the oxygen supply and demand bidirectional mapping diagram in this embodiment includes an oxygen supply diagram, an oxygen demand diagram, supply and demand control nodes, and a medical system interface; It should be further explained that the oxygen supply map in this embodiment is constructed by combining the connection relationships between each oxygen supply node and oxygen supply equipment with a graph algorithm; wherein, each oxygen supply node corresponds one-to-one with each oxygen supply interface; the oxygen demand map is constructed by each target to be supplied with oxygen label; the supply and demand control node is connected to each node in the oxygen supply map, and is used to obtain the real-time adjustable amount and total adjustable amount of oxygen for each oxygen supply interface, as well as the predicted real-time adjustable amount and total adjustable amount of oxygen within a future preset time length; the supply and demand control node is connected to the medical system interface, and is used to obtain the real-time oxygen demand index and the disease type and severity of each target to be supplied with oxygen; the supply and demand control node is connected to each node in the oxygen demand map, and is used to allocate the standard supply oxygen concentration and standard supply oxygen flow to each target to be supplied with oxygen according to the corresponding disease type and severity. It should be further explained that the specific reason for constructing the two-way oxygen supply and demand mapping map in this embodiment is to establish a digital twin model that can integrate physical resources, medical needs and intelligent control center, so as to realize panoramic, dynamic and precise management of the hospital oxygen supply system. By abstracting each oxygen supply interface as a node in the supply graph and mapping their connections, the system can track and predict the resource distribution and capacity changes of the entire oxygen supply network in real time. Simultaneously, by abstracting each patient as a node in the demand graph, individualized medical needs can be structurally represented. The core supply and demand control node, acting as the central hub of the bidirectional mapping, connects all supply nodes to aggregate resource data and connects to the medical system interface to import clinical needs. It then completes the targeted allocation of resources to patients by connecting all demand nodes. This design principle ensures that the real-time adjustable supply, individualized oxygen standards on the demand side, and disease-priority-based scheduling logic can be integrated, calculated, and optimized through simulation within a unified graph model. This provides a complete, consistent, and operable data foundation for subsequent prediction, decision-making, and anomaly localization, ultimately supporting closed-loop intelligent operation and maintenance from resource awareness to precise delivery.

[0021] It should be further explained that the supply and demand bidirectional forecasting model in this embodiment includes a supply forecasting sub-model, a demand forecasting sub-model, and a supply and demand discrimination layer; it should also be explained that the supply forecasting sub-model in this embodiment is built into the connection between the oxygen supply diagram and the supply and demand control node, and based on the real-time adjustable resource index, predicts the changing trend of the real-time adjustable oxygen quantity of each oxygen supply interface and the total adjustable oxygen quantity of the system within a preset time period; the construction of the supply forecasting sub-model includes the following steps: S201. Based on the historical time-series data of the real-time adjustable resource indicators, perform time-series feature engineering; specifically, for the historical sequence of adjustable oxygen quantity for each oxygen supply interface, extract feature vectors including lag value, sliding window mean, sliding window standard deviation and trend slope according to a preset time window length, and construct supervision labels corresponding to future preset time points to form an interface-level training sample set.

[0022] S202. Based on the principle of ensemble learning and the random forest algorithm, an initial supply prediction model is constructed. The model is composed of multiple decision trees as base learners. Each decision tree obtains a subset of samples through bootstrapping during training and grows by randomly selecting some features when splitting nodes to calculate the optimal split point.

[0023] S203. Based on the interface-level training sample set, the initial supply prediction model is trained in parallel; a recursive partitioning method is used to construct each decision tree by minimizing the impurity index of the nodes, and the prediction results of all decision trees are aggregated to output the final prediction value; this process is repeated until all decision trees in the forest are trained and a fully trained interface-level supply prediction model is obtained.

[0024] S204. Based on the fully trained interface-level supply prediction model, input the latest real-time adjustable resource index time series data of each oxygen supply interface after undergoing the same feature engineering processing, perform inference calculation, and obtain the predicted value of the real-time adjustable oxygen quantity of each oxygen supply interface at each target time point within a preset time length in the future.

[0025] S205. Based on the connection relationship and resource aggregation logic of the oxygen supply diagram, sum the predicted adjustable oxygen amount of all oxygen supply interfaces at the same target time point to obtain the predicted value of the total adjustable oxygen amount of the system at each time point within a future preset time length.

[0026] S206. Based on the predicted value of each oxygen supply interface and the predicted value of the total adjustable oxygen amount of the system, integrate and generate a report on the changing trend of the total adjustable oxygen amount of the system and an independent changing trend vector of each interface, and complete the construction and output of the supply prediction sub-model.

[0027] It should be further explained that the demand prediction sub-model in this embodiment is built into the connection between the medical system interface and the supply and demand control node. Based on the real-time oxygen demand index and the real-time oxygen priority index, it predicts the changing trend of the total oxygen demand target and the standard oxygen flow rate per unit time for all oxygen-supply targets within a preset time period, and aggregates them into the total oxygen supply demand changing trend. The specific implementation process includes: S211. Based on the medical system interface, obtain a fusion data sequence of historical real-time oxygen demand indicators and time-seriesd real-time oxygen priority indicators for all targets to be supplied with oxygen; the fusion data sequence includes the total oxygen demand target and standard oxygen flow rate per unit time for each target at consecutive historical time points.

[0028] S212. Based on the time series feature construction method, the fused data sequence is preprocessed; specifically, the data is aligned and organized according to the patient dimension, and a supervised learning sample set is constructed with sequence values ​​within the historical time window as features and sequence values ​​within a preset future time length as supervision labels.

[0029] S213. Based on the principle of recurrent neural networks and the structure of long short-term memory units, an initial demand prediction model is constructed. The model consists of an input layer, at least one LSTM hidden layer, and a fully connected output layer connected sequentially. The LSTM hidden layer contains input gates, forget gates, cell states, and output gate mechanisms to learn long-term dependencies.

[0030] S214. Based on the supervised learning sample set, the initial demand prediction model is trained in a supervised manner; the time series backpropagation algorithm is used to minimize the mean squared error loss function between the model prediction output and the true label, and the adaptive moment estimation algorithm is used to iteratively update the model weight parameters until the model loss converges, thereby obtaining a fully trained patient-level demand prediction model.

[0031] S215. Based on the fully trained patient-level demand prediction model, input the latest real-time fusion data sequence for all targets to be supplied with oxygen, perform inference calculations in parallel, and obtain the predicted value of the total oxygen demand target and the standard oxygen flow rate per unit time at each time point within a preset time period for each target.

[0032] S216. Based on resource aggregation logic, sum the predicted total oxygen demand of all targets to be supplied with oxygen at the same time point to obtain the change sequence of the total oxygen demand target; sum the predicted standard oxygen flow per unit time of all targets to be supplied with oxygen at the same time point to obtain the change sequence of the total standard oxygen flow of the system; finally, integrate and generate a report on the trend of total oxygen demand change, and complete the construction and output of the demand forecasting sub-model.

[0033] It should be further explained that the supply and demand discrimination layer in this embodiment is connected to the supply forecasting sub-model, the demand forecasting sub-model, and the enhanced allocation and control model, respectively. It is used to compare and detect conflicts between the changing trend of the total available oxygen supply and the changing trend of the total oxygen supply demand in real time. Based on preset discrimination thresholds and matching rules, it identifies potential oxygen supply gaps or resource redundancy risks and their corresponding timestamps within a preset time period, generates a supply-demand imbalance early warning signal, and inputs it into the enhanced allocation and control model. For example, this embodiment assumes that the supply and demand discrimination layer receives the changing trend of the total available oxygen supply in the system over the next six hours from the supply forecasting sub-model, which shows that the available supply will decrease to 500 liters per minute in the third hour; simultaneously, it receives the changing trend of the total demand for the same period from the demand forecasting sub-model, which shows that the total demand will rise to 580 liters per minute in the third hour. The supply and demand discrimination layer compares these two sets of time-series data in real time and, based on preset discrimination thresholds, for example, when the ratio of supply to demand exceeds 10%, it is determined to be a potential supply gap. After conflict detection, a supply gap of 80 liters per minute was identified in the third hour, along with the specific timestamp. Based on this, the supply and demand discrimination layer immediately generates a supply and demand imbalance early warning signal that includes the gap quantification value, the expected occurrence timestamp, and the type of imbalance. This signal is then input into the enhanced allocation and control model in real time to drive it to generate resource allocation or demand intervention strategies in advance.

[0034] It should be further explained that the specific implementation process of real-time comparison, conflict detection, and risk identification in the supply and demand discrimination layer of this embodiment includes: S221. Based on the trend sequence of the total adjustable oxygen quantity output by the supply prediction sub-model and the trend sequence of the total oxygen demand output by the demand prediction sub-model, extract their data points respectively; each sequence is indexed by a future timestamp at a fixed time interval, and each timestamp corresponds to a predicted value, thus forming two discrete time sequences of equal length and synchronized time base.

[0035] S222. Based on the linear interpolation algorithm, data point alignment processing is performed on two discrete time series. Specifically, if there are gaps in the data points of the two series at the same time stamp due to the slight difference in the output frequency of the prediction model, linear interpolation calculation is performed using adjacent known data points to fill the gaps, ensuring that the supply series and the demand series have unique and comparable prediction values ​​at each standardized future time stamp.

[0036] S223. Based on element-wise vector operation, for each standardized timestamp, the predicted value of the total available oxygen in the system is subtracted from the predicted value of the total oxygen demand trend to obtain a supply-demand difference sequence representing the supply and demand balance at each future time point.

[0037] S224. Perform automated conflict detection based on a preset set of discrimination thresholds; set a supply gap threshold, specifically a negative value, and set a resource redundancy threshold, specifically a positive value; the system traverses the supply-demand difference sequence, and when the difference at a certain timestamp is less than the supply gap threshold, the system automatically identifies and marks it as a potential supply gap event point; when the difference is greater than the resource redundancy threshold, the system automatically identifies and marks it as a potential resource redundancy event point.

[0038] S225. Risk confirmation and classification are performed based on the sliding window detection method and rule reasoning engine. The system takes each marked event point as the center and expands a preset time window forward and backward to check the number of consecutive time points marked as the same type within the window. If the consecutive number meets the minimum number of continuous points defined in the duration rule, it is confirmed as a valid continuous risk event. At the same time, the slope of the supply and demand difference sequence within the window is calculated. If the absolute value of the slope exceeds the threshold defined in the slope rule, the urgency of the risk event is divided into high, medium and low levels according to the preset standard.

[0039] S226. Based on the confirmed persistent risk event, extract its attributes; the system maps the start and end timestamps of the risk event from the standardized timestamp sequence according to the start and end positions of the time window; and finds the point with the largest absolute value of the supply-demand difference within the time window and maps it as the expected peak timestamp.

[0040] S227. Based on the information encapsulation protocol, a structured supply and demand imbalance early warning signal is generated. The system quantifies the risk type and risk intensity into the absolute value of the supply and demand difference, the duration of the risk, a set of key timestamps including the start peak and end times, and the urgency level, and encapsulates them into a standard format data object. The data object is then transmitted to the message interface of the enhanced allocation and control model in real time.

[0041] It should be further explained that the preset hierarchical mapping in this embodiment is constructed by combining the disease type, the severity of each disease type, the standard oxygen supply concentration under each disease type and the corresponding disease severity with the hierarchical oxygen supply standard interval using an association analysis algorithm; the hierarchical oxygen supply standard interval is constructed by the length of each oxygen supply time interval and the corresponding standard oxygen supply flow rate.

[0042] It should be further explained that the specific construction steps of the preset hierarchical mapping in this embodiment include: S231. Based on clinical treatment guidelines, expert consensus, and historical electronic medical record data, collect and organize different disease types and their corresponding severity grading standards; at the same time, based on the above medical literature and real-world data, extract the recommended standard oxygen supply concentration range and typical oxygen supply flow rate range for each disease type under each severity grade.

[0043] S232. Construct mapping rules based on association analysis algorithms; specifically, use the Apriori algorithm or FP-Growth algorithm to perform frequent itemset mining and association rule analysis on the sorted data of disease type, severity, concentration range and flow rate range, and find the rule combinations with strong correlation between disease type, severity and oxygen supply parameters as the initial mapping rule set.

[0044] S233. Based on medical expert review and clinical verification, the initial mapping rule set is reviewed and revised; medical experts confirm, adjust or supplement the rules mined by the algorithm based on clinical experience and the latest evidence to ensure the clinical rationality and safety of each mapping rule, thus forming the final mapping rule library.

[0045] S234. Based on the final determined mapping rule base, define a clear standard oxygen supply concentration value or a narrow range for each disease type and each severity level; at the same time, based on the division of typical treatment stages, define several continuous oxygen supply time intervals, and set a specific standard oxygen supply flow rate value for each time interval.

[0046] S235. The definition results obtained in S234 are structured and stored to construct a preset hierarchical mapping knowledge base. The knowledge base uses the disease type as the primary key and the severity level as the secondary index. Its entries include the standard oxygen supply concentration and a hierarchical oxygen supply standard interval sequence composed of multiple oxygen supply time interval lengths and standard oxygen supply flow rate pairs.

[0047] For example, this embodiment assumes that the disease type "acute exacerbation of chronic obstructive pulmonary disease" is classified into three severity levels: "mild," "moderate," and "severe." The pre-constructed hierarchical mapping will include the following entries: When the disease type is "acute exacerbation of chronic obstructive pulmonary disease" and the severity is "moderate," the standard oxygen supply concentration is defined as 28% to 35%, and the hierarchical oxygen supply standard range is defined as follows: during the initial 0-2 hours of treatment, the standard oxygen supply flow rate is 2 liters / minute; during the subsequent 2-24 hours, the standard oxygen supply flow rate is adjusted to 1.5 liters / minute. This entry will be stored in a structured form in a knowledge base for the system to directly query and match when generating a supply and demand allocation response strategy.

[0048] It should be further explained that the process of obtaining the supply and demand allocation response strategy in this embodiment includes: S241. Collect real-time oxygen demand indicators and real-time oxygen priority indicators through the configured medical system interface, and extract the hierarchical mapping features corresponding to each oxygen supply target. S242. Based on the hierarchical mapping features corresponding to each oxygen supply target, obtain the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate of each oxygen supply target under the current hierarchical mapping. It should be further noted that in real-world scenarios where clinical oxygen supply resources are scarce or instantaneous allocation capacity is limited, the idealized standard oxygen supply scheme may not be fully met. Therefore, the design intent of the minimum safe standard oxygen concentration and minimum safe standard oxygen flow rate for each target to be supplied under the current hierarchical mapping in this embodiment is to construct a dynamic degradation control mechanism based on oxygen supply safety benchmarks. The core of this mechanism is that after matching each target to be supplied with an idealized hierarchical mapping standard corresponding to its adaptation characteristics, the system does not fix the ideal supply standard, but further analyzes the minimum safe supply threshold that conforms to the oxygen supply safety specifications. The minimum safe oxygen supply concentration and the minimum safe oxygen flow rate together constitute the safety constraint baseline for supplying oxygen to the target. When real-time available oxygen supply resources cannot meet the ideal supply needs of all targets requiring oxygen supply, the allocation and control model, based on this safety constraint baseline and combined with preset real-time oxygen supply priority indicators, temporarily adopts the corresponding minimum safe supply standard for oxygen supply targets with lower priority levels and adaptability to low oxygen supply levels, provided that the minimum safe oxygen concentration and minimum safe oxygen supply flow rate requirements of each target under the current hierarchical mapping are met. Thus, under the constraint of limited overall oxygen supply resources, the key oxygen supply resources are prioritized for allocation to high-priority targets, ultimately achieving the safe and optimal operation of the system under non-ideal operating conditions and the emergency optimized allocation of oxygen supply resources.

[0049] It should be further explained that, in this embodiment, based on the preset hierarchical mapping knowledge base, entries that precisely match the adaptation features and levels of the current oxygen supply target are extracted to obtain the corresponding standard oxygen supply concentration value and hierarchical oxygen supply standard interval data. Relying on the associated oxygen supply safety threshold knowledge base, the minimum oxygen supply concentration value that matches the adaptation feature and level and meets the oxygen supply safety specifications is obtained through querying or calculation. This value is defined as the minimum safe standard oxygen supply concentration for the target. Simultaneously, based on the hierarchical oxygen supply standard interval, the minimum standard oxygen supply flow rate within the entire time interval is extracted, or the standard flow rate is calculated according to a preset degradation ratio rule. The calculated value is compared with the basic oxygen supply adaptation flow rate corresponding to the adaptation feature, and the higher of the two is selected as the minimum safe standard oxygen supply flow rate to meet basic oxygen supply requirements. Finally, the minimum safe standard oxygen supply concentration and the minimum safe standard oxygen supply flow rate are bound together as the safety benchmark parameter for the target oxygen supply under the current adaptation state. The oxygen supply safety threshold knowledge base in this embodiment relies on industry-standard oxygen supply safety, oxygen supply parameter research data, and historical successful oxygen supply adaptation case data. Through expert review, it summarizes and quantifies the oxygen supply adaptation tolerance thresholds corresponding to different adaptation characteristics and levels of the target oxygen supply, thereby pre-constructing a mapping table between adaptation characteristics and the minimum safe standard oxygen supply concentration. This embodiment pre-sets a degradation ratio rule. Based on emergency control strategies for oxygen supply resources and practical data on oxygen supply adaptation, it pre-sets an adjustment gradient for standard oxygen supply parameters through an expert review mechanism. For example, this rule can be defined as follows: under emergency operation conditions, the standard flow rate can be reduced sequentially by a maximum ratio of 20% and 40%, and the reduced flow rate value must not be lower than the basic oxygen supply demand flow rate corresponding to the adaptation characteristic. This rule has been coded and converted into executable logical judgment statements before system deployment.

[0050] S243. Collect the trend of oxygen demand changes at different concentrations within a preset historical time period, and combine it with a time series analysis algorithm with seasonal analysis characteristics to construct seasonal oxygen demand change curves at different concentrations and corresponding demand change functions. It should be further explained that the motivation for combining this embodiment with a time series analysis algorithm with seasonal analysis features is that: oxygen demand data of different concentrations within a preset historical time period are easily affected by factors such as changes in environmental temperature and humidity caused by seasonal changes, adjustments in industry production cycles, and fluctuations in residential oxygen demand, exhibiting significant seasonal cyclical variation patterns. Traditional time series analysis algorithms, due to insufficient consideration of seasonal characteristics, are prone to fitting biases and insufficient prediction accuracy in oxygen demand change trends. In order to accurately capture the inherent patterns of oxygen demand changes with the seasons and improve the scientificity and accuracy of demand change representation, it is necessary to introduce a time series analysis algorithm with seasonal analysis features. The specific technical means for constructing seasonal oxygen demand change curves of different concentrations and corresponding demand change functions is as follows: firstly, the collected historical data... The original data on oxygen demand at different concentrations over a given time period are preprocessed, including data cleaning, outlier removal, missing value imputation, and data standardization. Then, a time series analysis algorithm with seasonal characteristics is used to analyze the preprocessed data, separating trend, seasonal, and random components to clarify the seasonal variation cycle and magnitude of oxygen demand at different concentrations. Subsequently, based on the separated seasonal and trend components and corresponding concentration dimensions, nonlinear fitting methods (such as polynomial fitting and spline fitting) are used to construct seasonal oxygen demand variation curves for each concentration. Finally, combining the curve characteristics and parameters output by the time series analysis algorithm, a multivariate demand variation function incorporating time, seasonal, and concentration variables is constructed to quantitatively characterize the seasonal variation patterns of oxygen demand at different concentrations. The preferred time series analysis algorithm with seasonal characteristics is the STL time series decomposition algorithm or the SARIMA algorithm. S244. The seasonal oxygen demand variation curves and corresponding demand variation functions are embedded into the demand prediction sub-model. Combined with real-time collected oxygen demand indicators, the trend of total oxygen demand variation within a preset time period is predicted. The total oxygen demand variation trend includes the trend of total oxygen demand variation for all oxygen targets at each oxygen concentration, and the trend of the number of oxygen targets for each disease type and its corresponding severity. The specific implementation steps include: First, inputting the current time point into the seasonal demand variation function, calculating the predicted basic demand value for each oxygen concentration within the preset time period, and generating the basic demand. The system first calculates the total oxygen demand rate for each oxygen concentration using a curve. Secondly, based on real-time oxygen demand indicators, it calculates the current total demand rate for all oxygen targets at each oxygen concentration. Combining this with the basic demand curve, the predicted values ​​are dynamically corrected and fused using a Kalman filter algorithm, ultimately generating a trend of total oxygen demand for each oxygen concentration. Simultaneously, based on the disease type and severity information in the real-time oxygen demand indicators, it calculates the number of oxygen targets for each type and level in real-time, using this as the initial state. Combined with historical patient number change patterns, it uses time-series extrapolation to predict future changes in the number of targets, generating a trend of the number of oxygen targets for each disease type and corresponding severity. In this embodiment, the trend of total oxygen demand for each oxygen concentration is used to accurately calculate the resource gap or redundancy at each concentration level in the enhanced allocation and control model, providing direct quantitative basis for oxygen mixing ratio allocation and interface-level flow setting. The trend of the number of oxygen targets for each disease type and corresponding severity is used to dynamically adjust the weight distribution of the preset hierarchical mapping and predict the size changes of different priority patient groups, providing decision support for the system to implement priority scheduling strategies and emergency resource allocation plans in advance. For example, this embodiment assumes that during the winter forecast period, the system, through an embedded seasonal demand function, predicts that the total demand for 90% oxygen concentration will increase by 15% in the next 72 hours; simultaneously, real-time indicators show an increase in the number of patients with severe pneumonia. Based on this, the system generates a report on the trend of total oxygen demand changes, clearly indicating that the hourly demand for 90% oxygen concentration will peak at the 48th hour, increasing by 18% compared to the baseline; and the number of patients with "severe pneumonia - critical" level is expected to increase by 25% during the same period. Based on this, the enhanced allocation and control model will implement strategies in advance, prioritizing and locking in 90% oxygen concentration resources for patients with severe pneumonia, while dynamically adjusting the allocation ratio of medium-concentration oxygen, and issuing early warnings to maintenance personnel regarding equipment capacity increases, thereby ensuring supply and demand balance and treatment priority during peak periods.

[0051] S245. Collect data on the total oxygen supply variation over a time period equal to the oxygen demand, and combine this data with a time series analysis algorithm featuring seasonal characteristics to construct seasonal curves showing the total oxygen supply variation at different concentrations, as well as curves showing the supply variation at each interface. The core motivation for this step is to identify and quantify the regular fluctuations on the supply side. Similar to the demand side, oxygen supply capacity is also affected by seasonal or cyclical factors such as equipment operating efficiency, preventative maintenance cycles, and the stability of external gas supply. By constructing these curves, the system can model the key understanding that "supply capacity also fluctuates cyclically," thereby enabling more accurate matching calculations in two-way supply and demand forecasting, early detection of supply-demand imbalances exacerbated by seasonal declines in supply capacity, and providing a more forward-looking and realistic decision-making basis for strengthening the allocation and control model. Ultimately, this improves the robustness and reliability of the entire oxygen supply system throughout the year.

[0052] It should be further explained that the process of constructing the total oxygen supply variation curve with seasonally varying concentrations and the supply variation curve for each interface in this embodiment includes: Based on the original historical supply data, multi-dimensional data normalization is performed. Specifically, the total supply reading corresponding to each oxygen concentration at each historical time stamp is extracted to form a concentration-dimensional time series. At the same time, the independent supply reading of each physical supply interface at the same time stamp is extracted to form an interface-dimensional time series. For missing values ​​in the series, time series linear interpolation is used to fill them in. For outliers that are significantly outside the reasonable range, a threshold method based on statistical distribution is used for identification and correction. Finally, a normalized time series dataset with strictly aligned time stamps and complete data is obtained.

[0053] Based on the time series decomposition algorithm, the constituent components of each series are separated. Specifically, for each normalized supply total sequence of each concentration level and each interface supply sequence, the STL seasonal decomposition algorithm is applied. Through local weighted regression iterative loop, each original sequence is decomposed into long-term trend component, seasonal periodic component and residual random noise component, and the long-term trend component and seasonal periodic component are extracted for subsequent modeling.

[0054] Based on the periodic function fitting algorithm, a mathematical model of seasonal fluctuations is established. Specifically, for each extracted seasonal periodic component, the Fourier series expansion method is used for fitting, and the optimal fundamental and harmonic coefficients are solved by the least squares method to generate a parameterized mathematical expression with time as the independent variable that can accurately describe the seasonal fluctuation pattern.

[0055] Based on the component reconstruction algorithm, a complete prediction curve is generated. Specifically, the parameterized seasonal mathematical expression is linearly superimposed with the trend function fitted by the long-term trend component extracted in the second step to reconstruct a complete total supply change curve and interface supply change curve that can simultaneously reflect long-term changes and seasonal fluctuations.

[0056] Based on the parameterized storage protocol, the model output is solidified. Specifically, the key parameters of each reconstructed curve, including the trend function type and coefficients, the Fourier series coefficients of the seasonal expression, and the cycle length, are persistently stored in a structured data format, thereby completing the mathematical description and encapsulation of the historical supply pattern.

[0057] S246. The total supply variation curves of oxygen with different seasonal concentrations and the supply variation curves of each interface are embedded into the supply prediction sub-model to predict the total supply trend of each concentration of oxygen and the supply variation trend of each interface within the same preset time period in the future. The motivation for embedding the total supply variation curves of oxygen with different seasonal concentrations and the supply variation curves of each interface into the supply prediction sub-model in this embodiment is to overcome the shortcomings of traditional prediction methods that rely only on instantaneous data and short-term history. By systematically quantifying and introducing long-term patterns of supply capacity fluctuations with seasons, maintenance cycles, etc., the scientificity and foresight of medium and long-term predictions are significantly improved. The core purpose of this step is to provide the model with a high-precision baseline forecast. The supply forecast sub-model will use this baseline as a basis to dynamically correct the forecast by integrating real-time adjustable resource indicators and equipment operation anomaly indicators. This will output the trend of the total supply of the first concentration of oxygen within a preset time period, which reflects both long-term patterns and real-time conditions, as well as the trend of the supply at each interface. These trends are the key basis for the supply and demand discrimination layer to conduct forward-looking conflict detection and identify resource gaps caused by the periodic decline in supply capacity in advance. This will drive the enhanced allocation and control model to formulate cross-cycle preventive resource allocation strategies and equipment maintenance plans, ultimately achieving stable and reliable operation and optimal resource allocation of the supply system on an annual scale.

[0058] S247. Based on the predicted trend of total oxygen demand over a future preset time period, the trend of total available oxygen at each concentration over the same preset time period, and the trend of supply at each interface, the allocation time point when the total target oxygen demand exceeds the real-time available resource index is obtained. The motivation for this embodiment is to achieve a precise conversion from trend prediction to specific scheduling actions. Simply knowing that there is a risk of future supply-demand imbalance is insufficient; the scheduling system must know the exact time point at which the risk becomes urgent in order to activate contingency plans in a timely and effective manner before that point, such as calling backup gas sources, adjusting oxygen supply plans for non-emergency patients, or triggering equipment capacity expansion procedures. By accurately locating the allocation time point through the above steps, a clear decision-making trigger opportunity is provided for strengthening the allocation and control model, transforming forward-looking warnings into executable scheduling instructions, thereby seizing the initiative in handling the situation and preventing the system from falling into a passive response when an imbalance occurs.

[0059] It should be further explained that the process of obtaining the allocation time point where the total oxygen demand target is greater than the real-time adjustable resource index in this embodiment includes: S2471. Based on the trend of total oxygen demand change, extract the predicted total demand target value at each standard timestamp within a preset time period in the future to form a demand time series vector.

[0060] S2472. Based on the total supply trend of oxygen at each concentration, extract the predicted total supply value at each corresponding standard timestamp within the same future preset time length, and sum the predicted total supply values ​​of all concentration levels to form the prediction time series vector of the total adjustable resource index of the system.

[0061] S2473. Based on the time series comparison algorithm, the demand time series vector is compared with the predicted time series vector of the total available resources index of the system on a time-stamp basis, and the difference between the predicted demand value and the predicted supply value at each time stamp is calculated to generate a supply-demand gap time series vector.

[0062] S2474. Based on a preset supply gap discrimination threshold, traverse the time series vector of the supply-demand difference to identify all time points where the difference is greater than zero and its absolute value exceeds the discrimination threshold.

[0063] S2475. Based on the continuity rule, the identified time points are clustered and merged. Several consecutive time points that are identified as supply gaps are merged into a continuous potential gap period, and the start timestamp of this period is used as the key allocation time point output.

[0064] For example, by comparing the forecast curves, the system finds that at the timestamp of "the 36th hour in the future", the total demand forecast is 580 liters / minute, while the total supply forecast is only 520 liters / minute, a difference of 60 liters / minute, which exceeds the preset gap threshold of 50 liters / minute. At the same time, four consecutive time points from the 35th to the 38th hour are identified as gaps. Therefore, the system determines the starting timestamp of the 35th hour as the key allocation time point that requires the activation of the proactive allocation strategy.

[0065] S248. Based on the total supply trend of each oxygen concentration and the supply trend of each interface after the allocation time point, combined with the total demand trend of oxygen supply within the same time length, a multi-objective algorithm is used to solve the multi-objective oxygen supply allocation by combining the real-time oxygen supply priority index with the first optimization objective and the first optimization constraint, and the allocation solution is obtained. The first optimization constraint is: the real-time oxygen supply concentration and real-time oxygen supply flow rate corresponding to each oxygen supply target are greater than or equal to the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate corresponding to the current hierarchical mapping. The first optimization objective is: under the condition of satisfying the first optimization constraint, the supply satisfaction rate is equal to 1 and the oxygen supply risk of all oxygen supply targets is minimized. It should be further noted that in this embodiment, the supply satisfaction rate is prioritized to be equal to 1, and then the oxygen supply risk of all oxygen supply targets is minimized. It should be further explained that the specific process of solving the multi-objective oxygen supply allocation in this embodiment includes: Based on the assessment results of the disease type and severity of each target in the real-time oxygen supply priority index, a specific, numerical priority weight coefficient is assigned to each target by querying a preset priority-weight mapping table. This mapping table is established based on clinical consensus to ensure that the weight coefficient increases strictly monotonically with the increase of disease severity, and all coefficients are normalized so that their sum is one for subsequent weighted calculation.

[0066] Based on operations research modeling, a multi-objective optimization model is defined. The model's decision variable is a three-dimensional tensor, whose dimensions correspond to all discrete time points within a predetermined future time period, all oxygen supply interfaces, and all targets to be supplied with oxygen. Each element in the tensor is a non-negative real number, representing the oxygen flow rate allocated from a specific interface to a specific target at a specific time point. The primary constraint is expressed as a system of mathematical inequalities: for any target to be supplied with oxygen at any time point, the weighted average concentration of the allocated oxygen flow rate must not be lower than its minimum safe standard oxygen supply concentration, and the total allocated oxygen flow rate must not be lower than its minimum safe standard oxygen supply flow rate. The core optimization objective functions are defined as follows: The first objective function is the overall system supply satisfaction rate, which is the normalized value of one minus the total deviation between the standard demand and the actual allocation of all oxygen supply targets at all time points. The optimization direction is to maximize this function value and infinitely approach one. The second objective function is the comprehensive oxygen supply risk, which is the sum of the individual oxygen supply risk values ​​of all oxygen supply targets at all time points. The individual oxygen supply risk value is equal to the negative deviation between the standard oxygen supply concentration and the actual weighted average concentration, the negative deviation between the standard oxygen supply flow rate and the actual total flow rate, and the cumulative value after being weighted by the priority weight coefficient and multiplied by the time penalty factor. The optimization direction is to minimize this function value.

[0067] The solution is based on the second-generation non-dominated sorting genetic algorithm in multi-objective evolutionary algorithms. Specifically, the following steps are taken: First, an initial population is generated. Each individual in the population has a chromosome encoding a complete allocation scheme tensor. In each generation iteration, the algorithm performs the following operations: Tournament selection is performed on the parent population to retain better individuals; simulated binary crossover and polynomial mutation are performed on the selected individuals to generate the offspring population; the parent and offspring populations are merged, and a fast non-dominated sorting algorithm is used to divide all individuals in the merged population into multiple non-dominated levels based on the two objective function values; to maintain population diversity, the crowding degree of each individual within the same non-dominated level is calculated, i.e., the distance between its neighboring individuals in the objective function space; individuals are sorted according to the non-dominated level and the crowding degree, and individuals of a predetermined size are selected to form the new generation population. This iterative process continues until the preset maximum number of generations is reached, and the final population approximately constitutes the Pareto optimal solution set for the optimization problem.

[0068] Based on the multi-criteria decision-making method, the final implementation scheme is selected from the obtained Pareto optimal solution set. The specific values ​​of the two objective functions corresponding to each candidate scheme in the solution set are calculated. First, the schemes that make the value of the first objective function closest to 1 are selected as the candidate subset. Then, in this candidate subset, the values ​​of the second objective function of each scheme are compared, and the scheme with the smallest comprehensive oxygen supply risk value is selected. If there are multiple schemes with the same risk value, a fairness auxiliary index based on priority weight coefficient is further introduced for discrimination. Finally, a unique scheme is selected as the multi-objective oxygen supply allocation scheme to be implemented.

[0069] S249. Based on the allocation solution results, if there is an allocation solution result that satisfies the supply satisfaction rate of 1, then the standard supply oxygen concentration and the stratified oxygen supply standard range in the stratified mapping corresponding to each node in the oxygen demand diagram after the allocation time point are adjusted according to the current solution result. The adjusted stratified mapping is used as the new supply and demand allocation response strategy until the total oxygen demand target is less than or equal to the real-time adjustable resource index, and then the supply and demand allocation response strategy corresponding to the original stratified mapping is restored. In this embodiment, under a sudden surge in oxygen demand, when the number of targets awaiting oxygen supply increases dramatically, the centralized oxygen supply system will face extreme operational pressure: on the one hand, the demand for high-concentration oxygen resources from a large number of high-priority targets increases exponentially; on the other hand, external oxygen supply links are disrupted by objective factors, and the capacity of local oxygen production equipment reaches its threshold limit in a short period of time, resulting in a continuous and severe supply gap in oxygen resources. Under such extreme conditions, if the conventional high-standard oxygen allocation scheme is used, oxygen resources will be rapidly consumed in a short period of time, ultimately resulting in the inability to effectively meet the oxygen supply needs of subsequent targets. Therefore, the core design idea and emergency response logic of this mechanism is to activate a preset emergency control procedure based on oxygen supply safety benchmarks. The system first relies on the real-time monitoring module and resource prediction model to determine the degree of oxygen resource shortage and the duration of the shortage; then, based on the enhanced allocation control model and multi-objective optimization algorithm, it conducts a global reassessment and dynamic allocation of the oxygen allocation scheme for all targets awaiting oxygen supply. Under the premise of ensuring that the oxygen supply parameters of all targets awaiting oxygen supply are not lower than the minimum safe supply standard corresponding to their adaptation characteristics and level and in accordance with industry oxygen supply safety standards, for example, the oxygen supply concentration of some medium-level targets awaiting oxygen supply can be temporarily and standardizedly lowered from the optimal safe supply concentration to the minimum safe supply concentration that meets basic oxygen supply needs. This allows surplus high-concentration oxygen supply resources to be concentrated and prioritized for supply to the highest priority targets. This standardized temporary downgrade of oxygen supply standards essentially involves refined and quantitative allocation of oxygen supply resources within the bottom line of oxygen supply safety constraints. It can significantly reduce the overall consumption rate of oxygen supply resources, effectively extending the continuous oxygen supply duration of the system from hours to days or even several days. This provides an effective time window for emergency allocation of external oxygen supply resources, deployment of temporary oxygen preparation equipment, or alleviation of the imbalance between oxygen supply and demand. The system will continuously monitor key indicators of oxygen supply and demand. When it determines that external oxygen supply has stabilized or the imbalance between supply and demand has been alleviated, it will immediately and automatically complete the smooth rollback of the oxygen allocation plan for the entire target oxygen supply, restoring to the normal high-standard oxygen supply mode. Thus, under extreme conditions of oxygen shortage, the system achieves the goal of time-for-space regulation, maximizing the utilization efficiency of oxygen resources while ensuring the safety of oxygen supply for the entire target oxygen supply.

[0070] S250. When no allocation solution exists that satisfies a supply satisfaction rate of 1, a forward search is performed using the aforementioned allocation time point combined with the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate for each target to be supplied under the current hierarchical mapping, until an allocation solution exists that satisfies a supply satisfaction rate of 1. The last time point of the forward search is then used as the new allocation time point. Please refer to [link to previous section]. Figure 3The A to B section represents the length from historical time point A to the current time point B, the B to E section represents the length from the current time point B to the end of the future preset time length E, the D section represents the initial allocation time point, and the C section represents the time point corresponding to the allocation solution where the supply satisfaction rate is equal to 1 for the first time after the forward search, i.e., the new allocation time point. It needs further explanation that the specific motivation for this step is that when the system, through the supply and demand discrimination layer, identifies an absolute gap where the total oxygen demand exceeds the real-time available resources at a specific future allocation time, and a feasible solution with a supply satisfaction rate of one cannot be obtained through conventional multi-objective oxygen allocation at the current time or at that time, it indicates that the system is about to face a resource crisis that cannot be addressed through immediate optimization. To avoid being forced to execute suboptimal allocations that may endanger patient safety in emergencies, the system must activate a proactive and forward-looking resource pre-allocation mechanism. Its core strategy is: instead of passively waiting for the gap to occur, it looks back from the future allocation time to an earlier, relatively resource-sufficient current period, planning and reserving oxygen resources in advance. This ensures that, during the period before the gap occurs, through orderly allocation intervention, sufficient oxygen resources are accumulated and reserved for high-priority oxygen supply targets, enabling these resources to support and meet the critical needs at the future gap time. This mitigates the risk of an absolute gap in advance, providing a valuable response window for external resource allocation or system expansion. The specific implementation steps are as follows: First, the initial future allocation time point identified by the supply and demand discrimination layer is used as the starting point for backtracking search. Then, based on a preset fixed time step, the system iterates backwards towards the current time, generating a series of earlier candidate pre-allocation time points. For each generated candidate time point, the system performs the following operations: Based on the oxygen supply and demand bidirectional mapping, it obtains the latest predicted supply and demand trend data from the candidate time point to the future gap time point; then, based on the real-time oxygen supply priority indicators of all targets to be supplied, and with ensuring the standard supply demand of high-priority targets at the future gap time point as the core optimization objective, a resource reservation optimization model is reconstructed. The constraints of this model include that, from the candidate time point to the future gap time point, in each time period, the total available resources of the system, while meeting the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate of all targets to be supplied, must reserve the standard supply amount required by the designated high-priority targets at the future gap time point. The system calls the solution algorithm to verify the feasibility of this reservation model. This backtracking search process continues, with the system sequentially verifying each earlier candidate time point until a candidate time point that successfully passes the feasibility verification is found. This means that executing the reservation strategy at this point will not cause any violation of the minimum safety standard constraints during any period. Finally, the system officially determines the first candidate time point that meets the conditions as the new critical action point for immediately implementing the resource pre-allocation strategy, and injects the corresponding resource reservation scheme as a new supply and demand allocation response strategy that must be enforced in the current stage into the enhanced allocation control model for execution.

[0071] S251. Use the allocation solution results to adjust the hierarchical mapping after the new allocation time point as a new supply and demand allocation response strategy until the total oxygen demand target is less than or equal to the real-time available resource index, then restore the supply and demand allocation response strategy corresponding to the original hierarchical mapping.

[0072] It should be further explained that the specific implementation steps of feeding the positioning results back to the enhanced allocation and control model in this embodiment to adjust the supply and demand allocation response strategy in real time include: S301. Based on real-time data stream, obtain the specific indicator item that triggers the early warning; when the system detects that any value of the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter does not meet the preset constraint indicator value, immediately lock the abnormal indicator item, and simultaneously obtain the specific value, occurrence timestamp, and directly associated entity identifier of the abnormal indicator item, wherein the entity identifier is the target label to be supplied with oxygen or the oxygen supply interface number.

[0073] S302. Based on the topology of the oxygen supply and demand bidirectional mapping graph, obtain the graph nodes associated with the abnormal entities; if the abnormal entity is a target label for oxygen supply, then obtain the corresponding target node for oxygen supply from the oxygen demand graph; if the abnormal entity is an oxygen supply interface number, then obtain the corresponding oxygen supply node from the oxygen supply graph; this step determines the starting node for causal analysis in the mapping graph.

[0074] S303. Based on the Peter-Clark algorithm in graph theory and a pre-defined causal relationship rule base, the system performs root cause localization analysis on the mapping graph. Taking the starting node obtained in S302 as the center, the system uses the constraint-based Peter-Clark algorithm to explore all potential causal paths within a specific time window before the time of the anomaly. Simultaneously, it queries the causal relationship rule base pre-constructed through historical fault data mining and expert knowledge. This base defines the correspondence between various indicator anomaly patterns and specific node state anomalies, edge flow anomalies, or control logic anomalies in the graph. By matching the causal paths explored by the algorithm with the patterns in the rule base and calculating the confidence level, the system finally locates one or a few root cause nodes or key failed connection edges that are most likely to cause the current anomaly. Specifically, in this embodiment, when either the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter fails to meet the preset constraint value, after monitoring and triggering an early warning in step S3, this method performs anomaly root cause localization analysis based on the Peter-Clark algorithm in the oxygen supply and demand bidirectional mapping diagram. Taking the real-time oxygen supply flow rate of bed 12 interface continuously lower than the preset constraint value as an example, the method includes: First, taking the starting node associated with the abnormal indicator that triggered the warning as the center, specifically the demand node in the oxygen demand graph corresponding to the target oxygen supply that triggered the warning, i.e., the demand node corresponding to bed 12. In the bidirectional oxygen supply and demand mapping graph, all associated nodes within the K-order neighborhood of this starting node are extracted, where the K-order neighborhood is defined as the set of nodes connected by edges with a path length not exceeding K. These associated nodes include the upstream oxygen supply node, i.e., the supply node corresponding to the No. 3 manifold pressure reducing valve, other downstream oxygen demand nodes, and supply and demand control nodes.

[0075] Second, multivariate time-series operating parameters of these nodes at multiple consecutive sampling times prior to the occurrence of the anomaly are obtained from real-time databases and historical monitoring data. These operating parameters include real-time oxygen supply concentration, real-time oxygen supply flow rate, operating status of digital medical oxygen flow meter, adjustable oxygen supply volume, and standard oxygen supply concentration. A dataset to be analyzed is constructed based on this, where each row corresponds to a sampling time and each column corresponds to a specific operating parameter of a node.

[0076] Third, the dataset is input into the Peter-Clark algorithm, and the causal relationship between nodes is explored by the conditional independence test method based on the Fisher Z test: the Fisher Z test is used to determine whether two variables are conditionally independent given other variables. Starting from a completely undirected graph (i.e., a graph that assumes there are potential connections between all nodes), conditionally independent edges are gradually removed to obtain the skeleton of the undirected graph representing potential associations.

[0077] Fourth, orientation rules based on the Peter-Clark algorithm are used to determine the orientation of edges in the undirected graph skeleton. These orientation rules include V-structure orientation rules and Meek rules. V-structure orientation rules are used to identify structures with common child nodes and determine the direction of edges, while Meek rules are used to further propagate directional constraints, thereby generating a directed acyclic graph that reflects the anomaly propagation path. This directed acyclic graph is the candidate causal structure behind the current anomaly event.

[0078] Fifth, identify causal paths from the anomaly indicator node (i.e., the demand node for bed 12) to all possible root nodes in the directed acyclic graph. Potential root nodes are nodes with an in-degree of zero (i.e., nodes without a parent node) or nodes with significant temporal leadership; these nodes specifically correspond to a particular oxygen supply interface or oxygen supply equipment component. By traversing the directed acyclic graph, several candidate causal paths are obtained, starting from the anomaly indicator node and backtracking along directed edges to each root node.

[0079] Sixth, the system matches and calculates the confidence scores of the candidate causal paths identified by the algorithm with a pre-defined causal association rule base. This rule base, pre-defined through historical fault data mining and expert knowledge construction, defines the correspondence between various abnormal indicator patterns and specific node state anomalies, edge traffic anomalies, or control logic anomalies in the graph. Specifically, for each candidate causal path, the system extracts the state change patterns of each node on the path (e.g., the temporal change characteristics of node parameters), compares them with the abnormal patterns stored in the rule base, and calculates the matching score (i.e., confidence score) for each path. The higher the matching score, the higher the degree of agreement between the path and the known fault patterns.

[0080] Seventh, the path-end node with the highest confidence score that points to the supply source (i.e., the path-end node is a supply-side node) is identified as the root cause node of the anomaly. In this example, the unique identifier for locating the root cause node through the above analysis is the No. 3 manifold pressure reducing valve, whose anomaly is caused by diaphragm aging leading to unstable oxygen supply pressure. The system feeds this location result back to the enhanced allocation and control model to adjust the supply and demand allocation response strategy in real time, such as adjusting the opening of other valves or switching to a backup pressure reducing valve; at the same time, a warning is displayed on the interface to prompt maintenance personnel to replace the diaphragm. Thus, a closed-loop intelligent operation and maintenance process from anomaly monitoring, root cause location to strategy correction is completed.

[0081] S304. Based on the location analysis conclusions, generate a structured anomaly location and diagnosis report. This report must clearly state the unique identifier of the root node or edge located, determine the type of the root cause of the anomaly, such as equipment performance degradation of a specific oxygen supply interface, an unexpected surge in the demand parameters of a target to be supplied with oxygen, or a calculation deviation in the previous allocation strategy executed by the supply and demand control node on a specific path, and explain in detail which downstream demand nodes or upstream supply nodes in the graph are covered by its impact. Finally, based on the preset repair knowledge base, generate one or more targeted recommended control actions, such as switching the specified target to be supplied with oxygen to the backup supply path, or immediately performing flow calibration and compensation on a specific oxygen supply interface. Specifically, in this embodiment, when generating a structured anomaly location and diagnosis report, the determination of the root cause type of the anomaly includes the following situations: When the located anomaly root cause node is an oxygen supply node, the root cause type is determined to be equipment performance degradation of a specific oxygen supply interface, specifically manifested as the real-time adjustable oxygen quantity of the supply node being continuously lower than the predicted value output by the supply prediction sub-model, or the abnormal operation index of the digital display medical oxygen flow meter corresponding to the supply node exceeding a preset threshold; When the located anomaly root cause node is an oxygen demand node, the root cause type is determined to be non-standard demand parameters of a specific oxygen supply target. A surge in demand is expected, specifically manifested when the target total oxygen demand or standard oxygen supply flow rate in the real-time oxygen demand index of the demand node exceeds the range of standard supply parameters corresponding to its current hierarchical mapping. When the identified root cause node is a supply and demand control node, the root cause is determined to be a calculation deviation in the previous allocation strategy executed by the supply and demand control node on a specific path. This is specifically manifested as a systematic discrepancy between the allocation logic output by the supply and demand control node and the actual supply and demand data, i.e., the allocation instructions issued by the node do not match the actual output flow of downstream supply nodes or the actual consumption flow of upstream demand nodes. The system encapsulates the above-identified root cause types and their specific manifestations into the anomaly location and diagnosis report to guide the enhanced allocation control model in adjusting its strategy, and prompts maintenance personnel to perform corresponding repair operations through the display interface.

[0082] S305. The generated anomaly location and diagnosis report is fed back to the policy update interface of the enhanced allocation control model in real time. Based on the received report, the enhanced allocation control model first updates its internally maintained state space representation, that is, modifies or marks the state values ​​of the root node and affected node indicated in the report as anomalies, and incorporates this event as a significant negative reward signal into its immediate reward calculation. Subsequently, the model focuses its action space search range on the control action direction recommended in the report, and uses its built-in policy network to re-evaluate and generate a new set of candidate supply and demand allocation response strategies that have avoided known anomalies. After simulation evaluation, the model outputs the finally selected corrected supply and demand allocation response strategy, thereby completing the real-time policy adjustment closed loop based on causal location feedback.

[0083] For example, in this embodiment, during the real-time operation of the oxygen supply system, the simulation control module detected that the oxygen flow rate at the interface of bed 12 was consistently lower than the preset safety threshold of 2.5 liters per minute. According to the standard requirement for this bed's severe pneumonia classification, the flow rate should be 4 liters per minute, but the actual flow rate was only 1.8 liters per minute, resulting in a supply satisfaction rate of 0.45. The system immediately triggered an early warning. Using the Peter-Clark algorithm to trace the cause and effect in the oxygen supply and demand bidirectional mapping graph, the root cause of the anomaly was located as aging of the diaphragm of the pressure reducing valve at manifold 3, leading to outlet pressure fluctuations and causing unstable flow in the oxygen supply pipeline connected to this valve at bed 12. Upon receiving the positioning results, the enhanced allocation and control model immediately initiates a strategy adjustment process based on a discrete event simulation algorithm. This algorithm first constructs a virtual image of the current system, setting the initial state including the real-time adjustable quantities of all supply interfaces. For example, the current total oxygen output of manifold 3 is 120 liters per minute, but due to aging, the actual fluctuation range is 80 to 110 liters per minute. The demand of each bed is also set, such as bed 12 requiring 4 liters per minute, the adjacent bed 13 requiring 3 liters per minute, and bed 14 requiring 5 liters per minute, as well as the predicted supply and demand curve for the next 4 hours. Next, a set of simulation events is defined, including candidate actions such as switching bed 12 to the standby manifold 4, adjusting the output pressure of manifold 3 to a stable range, and two intervention combinations. Execution time and effect parameters are set for each action. For example, switching the standby valve is expected to take 15 seconds, and after the switch, bed 12 can obtain a stable flow rate of 3.8 liters per minute. The simulation engine advances virtual time in second-level steps, updating the system status at each time step, recording the actual oxygen concentration and flow rate, supply satisfaction rate, and oxygen supply risk indicators for each bed, and accumulating the overall resource utilization rate. After multiple simulation runs, the model compares the cumulative results under different strategies. For example, Scheme A, which only switches the standby valve, can restore the flow rate of bed 12 to 3.8 liters per minute and increase the supply satisfaction rate to 0.95, but the total output of the standby manifold increases from 200 liters per minute to 235 liters per minute, close to the upper limit, which may affect the sudden demand of other beds in the next 2 hours. Scheme B, which switches the standby valve and simultaneously reduces the output pressure of manifold 3 from 0.4 MPa to 0.35 MPa to suppress fluctuations, stabilizes the flow rate of bed 12 at 3.6 liters per minute, with a supply satisfaction rate of 0.9, while narrowing the output fluctuation of manifold 3 to 95 to 105 liters per minute, reducing total resource consumption, and decreasing the overall system oxygen supply risk value from 0.32 to 0.18. The discrete event simulation algorithm outputs scheme B as the optimal solution. The enhanced allocation and control model then uses it as a new supply and demand allocation response strategy and executes it. Simultaneously, the positioning result of the aging of the diaphragm of the No. 3 manifold pressure reducing valve and the pressure adjustment and switching operation to be executed are pushed to the display interface for real-time early warning. This realizes the closed-loop intelligent control of the entire process from anomaly monitoring, root cause location, simulation verification to strategy adjustment and early warning.

[0084] This application addresses the isolation and lag issues in monitoring oxygen supply parameters (concentration and flow rate). By defining a composite real-time oxygen supply quality evaluation index, integrating core dimensions such as oxygen concentration, flow rate, and supply satisfaction rate, a multi-dimensional dynamic evaluation system is established. This system can accurately capture parameter deviations caused by equipment operation fluctuations, pipeline transmission anomalies, or scheduling mismatches. More importantly, this method initializes a bidirectional oxygen supply and demand mapping topology, abstracting each oxygen supply interface in the physical scenario as a supply node and each target to be supplied as a demand node. Through supply and demand control nodes, bidirectional association and data exchange between nodes are achieved, constructing a complete digital twin model of the oxygen supply system. This model supports the system in real-time tracking of oxygen resource flow and the status of targets to be supplied from a panoramic and structured perspective. Combined with hierarchical mapping rules based on the adaptation characteristics and pre-defined levels of targets to be supplied, the system can accurately match differentiated standard oxygen supply schemes for each target. When oxygen supply resources are scarce, the minimum safety standard supply parameters obtained through dynamic analysis based on the oxygen supply safety threshold knowledge base can set an inviolable safety constraint bottom line for resource scheduling. This ensures that even in emergency situations, the oxygen supply level of each target to be supplied can be maintained above the oxygen supply safety benchmark, significantly improving the overall safety of oxygen supply. Simultaneously, it represents a fundamental upgrade from an experience-based oxygen allocation model to a precise and differentiated oxygen allocation model based on industry oxygen supply standards and adaptive data. Addressing the lack of foresight in traditional scheduling, this method innovates by constructing a two-way supply and demand forecasting model: the supply forecasting sub-model, combining real-time data with seasonal supply capacity change curves obtained through historical data analysis, can predict the impact of cyclical equipment efficiency decline and planned maintenance on long-term oxygen supply capacity; the demand forecasting sub-model embeds a seasonal demand change function, combining real-time oxygen supply priority indicators to predict the number and demand trends of different oxygen supply groups; the supply and demand discrimination layer, by comparing supply and demand forecast trends, identifies potential oxygen supply gaps or resource redundancy risks and corresponding timestamps in advance, generating forward-looking warnings and driving resource scheduling from passive response to proactive planning. After the warning is triggered, the enhanced allocation and control model, based on a multi-objective optimization algorithm, seeks the optimal allocation scheme with the goal of minimizing the overall supply satisfaction rate and the comprehensive oxygen supply risk, while meeting the minimum safety standards of all targets awaiting oxygen supply, thus balancing resource utilization efficiency and oxygen supply risk. In extreme cases where conventional optimization fails, the system initiates a backtracking search mechanism, looking back from future gap points to periods of sufficient resources to find opportunities for reservation strategies. By reserving resources in advance for high-priority targets awaiting oxygen supply, the system mitigates long-term crises. This "time for space" strategy significantly enhances the system's resilience in the face of absolute resource shortages, providing a window of opportunity for external resource allocation.

[0085] Example 2 Please see Figure 4Another embodiment of the present invention provides an oxygen supply and demand regulation and optimization system using a digital display oxygen flow meter, comprising: The data acquisition module is used to acquire real-time oxygen demand indicators, real-time oxygen priority indicators, real-time adjustable resource indicators, real-time oxygen supply quality indicators, and abnormal operation indicators of the digital medical oxygen flow meter. The enhanced allocation module is used to initialize a preset oxygen supply and demand bidirectional mapping diagram. It combines real-time oxygen demand indicators, real-time oxygen priority indicators, and real-time adjustable resource indicators with a preset supply and demand bidirectional prediction model and an enhanced allocation control model to obtain supply and demand allocation response strategies. The simulation control module is used to respond to the supply and demand allocation response strategy on the oxygen supply and demand bidirectional mapping map within the preset oxygen supply simulation control platform, and to monitor real-time oxygen supply quality indicators and abnormal operation indicators of the digital medical oxygen flow meter. When either the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter fails to meet the preset constraint index value, the Peter-Clark algorithm is used to analyze and locate the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter on the oxygen supply and demand bidirectional mapping map, and the location result is fed back to the enhanced allocation control model. Combined with the simulation algorithm, the supply and demand allocation response strategy is adjusted in real time, and the location result is simultaneously fed back to the configured display interface for real-time early warning display.

[0086] Example 3 An electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement an oxygen supply and demand regulation optimization method using a digital display oxygen flow meter.

[0087] A computer-readable storage medium storing computer instructions that, when executed, perform an oxygen supply and demand regulation optimization method using a digital oxygen flow meter.

[0088] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of the present invention without departing from the spirit and scope of the claims. All of these variations are within the protection scope of the present invention.

[0089] If the technical solution disclosed herein involves personal information, the product using this technical solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If the technical solution disclosed herein involves sensitive personal information, the product using this technical solution has obtained the user's separate consent before processing the sensitive personal information, and also meets the requirement of "express consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set up to inform users that they have entered the scope of personal information collection and that personal information will be collected. If an individual voluntarily enters the collection scope, it is deemed that they have agreed to the collection of their personal information; or on the personal information processing device, with clear signs / information informing users of the personal information processing rules, authorization is obtained from the individual through pop-up information or by asking the individual to upload their personal information; wherein, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

Claims

1. A method for optimizing oxygen supply and demand control using a digital display oxygen flow meter, characterized in that, include: Acquire real-time oxygen demand indicators, real-time oxygen priority indicators, real-time adjustable resource indicators, real-time oxygen supply quality indicators, and abnormal operation indicators of digital medical oxygen flow meters. Initialize a preset oxygen supply and demand bidirectional mapping diagram, input the real-time oxygen demand index, real-time oxygen priority index, and real-time adjustable resource index into the preset supply and demand bidirectional prediction model and enhanced allocation control model, and combine them with the preset hierarchical mapping to obtain the supply and demand allocation response strategy. On the oxygen supply and demand bidirectional mapping map within the preset oxygen supply simulation and control platform, the system responds to the supply and demand allocation response strategy and monitors real-time oxygen supply quality indicators and abnormal operation indicators of the digital medical oxygen flow meter. When either the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter fails to meet the preset constraint value, the Peter-Clark algorithm is used to analyze and locate the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter on the oxygen supply and demand bidirectional mapping map. The location result is then fed back to the enhanced allocation and control model, and the supply and demand allocation response strategy is adjusted in real time in conjunction with the simulation algorithm. Simultaneously, the location result is fed back to the configured display interface for real-time early warning display.

2. The oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in claim 1, characterized in that, The real-time oxygen demand index includes the target label corresponding to each oxygen supply target, the total oxygen demand target for each oxygen concentration, and the standard oxygen supply flow rate; the real-time oxygen supply priority index is constructed from the disease type and severity of each oxygen supply target; the real-time adjustable resource index is obtained by measuring the adjustable oxygen quantity under all oxygen supply interfaces corresponding to each oxygen concentration; the real-time oxygen supply quality index is constructed by combining the real-time oxygen supply concentration and the real-time oxygen supply flow rate to achieve the supply satisfaction rate; the supply satisfaction rate is constructed by multiplying the ratio of the real-time oxygen supply concentration to the standard oxygen supply concentration, the ratio of the real-time oxygen supply flow rate to the standard oxygen supply flow rate in the corresponding oxygen supply time interval, and the ratio of the actual total oxygen supply to the total oxygen demand target for each oxygen supply target. The abnormal operation index of the digital display medical oxygen flow meter is obtained by combining the operating parameters of each digital display medical oxygen flow meter with the abnormal frequency and lifespan change of the digital display medical oxygen flow meter through an evaluation algorithm with a built-in Bayesian function. It is used to characterize the probability of each digital display medical oxygen flow meter experiencing an abnormality or failure within a preset time period in the future.

3. The oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in claim 2, characterized in that, The oxygen supply and demand bidirectional mapping diagram includes an oxygen supply diagram, an oxygen demand diagram, supply and demand control nodes, and a medical system interface; The oxygen supply map is constructed using a graph algorithm, combining the connection relationships between each oxygen supply node and oxygen supply equipment. Each oxygen supply node corresponds one-to-one with each oxygen supply interface. The oxygen demand map is constructed from the labels of each target to be supplied with oxygen. The supply and demand control node is connected to each node in the oxygen supply map to obtain the real-time adjustable quantity and total adjustable oxygen quantity of each oxygen supply interface, as well as the predicted real-time adjustable quantity and total adjustable oxygen quantity within a predetermined future time period. The supply and demand control node is connected to the medical system interface to obtain real-time oxygen demand indicators and the disease type and severity of each target to be supplied with oxygen. The supply and demand control node is also connected to each node in the oxygen demand map to allocate the standard oxygen concentration and total standard oxygen flow required for each target to be supplied with oxygen based on its corresponding disease type and severity.

4. The oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in claim 3, characterized in that, The supply and demand two-way forecasting model includes a supply forecasting sub-model, a demand forecasting sub-model, and a supply and demand discrimination layer. The supply prediction sub-model is built into the connection between the oxygen supply diagram and the supply and demand control node. Based on the real-time adjustable resource index, it predicts the changing trend of the real-time adjustable oxygen quantity of each oxygen supply interface and the total adjustable oxygen quantity of the system within a preset time period. The demand forecasting sub-model is built into the connection between the medical system interface and the supply and demand control node. Based on the real-time oxygen demand index and the real-time oxygen priority index, it predicts the change trend of the total oxygen demand target and the standard oxygen flow rate per unit time of all oxygen-supply targets within a preset time period, and aggregates them into the total oxygen demand change trend.

5. The oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in claim 4, characterized in that, The supply and demand discrimination layer is connected to the supply forecasting sub-model, the demand forecasting sub-model, and the enhanced allocation and control model, respectively. It is used to compare and detect conflicts between the changing trend of the total adjustable oxygen quantity of the system and the changing trend of the total oxygen supply demand in real time. Based on the preset discrimination threshold and matching rules, it identifies potential oxygen supply gaps or resource redundancy risks and corresponding timestamps within a preset time period in the future, generates a supply and demand imbalance early warning signal, and inputs it into the enhanced allocation and control model. The preset hierarchical mapping is constructed by combining the disease type, the severity of each disease type, the standard oxygen supply concentration under each disease type and the corresponding disease severity with the hierarchical oxygen supply standard interval using a correlation analysis algorithm. The hierarchical oxygen supply standard interval is constructed by the length of each oxygen supply time interval and the corresponding standard oxygen supply flow rate.

6. The oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in claim 5, characterized in that, The process of obtaining the supply and demand allocation response strategy includes: The system collects real-time oxygen demand indicators and real-time oxygen priority indicators through the configured medical system interface, and extracts the hierarchical mapping features corresponding to each oxygen supply target. Based on the hierarchical mapping features corresponding to each oxygen supply target, the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate of each oxygen supply target under the current hierarchical mapping are obtained. Collect the trend of oxygen demand changes at different concentrations within a preset historical time period, and combine it with a time series analysis algorithm with seasonal analysis characteristics to construct seasonal oxygen demand change curves at different concentrations and corresponding demand change functions. The seasonal oxygen demand curves and corresponding demand change functions are embedded into the demand prediction sub-model. Combined with real-time collected oxygen demand indicators, the trend of total oxygen demand over a future preset time period is predicted. The total oxygen demand trend includes the total oxygen demand trend of all oxygen targets corresponding to each oxygen concentration, and the trend of the number of oxygen targets under each disease type and corresponding severity.

7. The oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in claim 6, characterized in that, The process of obtaining the supply and demand allocation response strategy also includes: By collecting data on the total oxygen supply variation over the same time period as oxygen demand and combining it with a time series analysis algorithm with seasonal analysis characteristics, we can construct seasonal curves of total oxygen supply variation at different concentrations and curves of supply variation at each interface. The total supply variation curves of oxygen with different seasonal concentrations and the supply variation curves of each interface are embedded into the supply prediction sub-model to predict the total supply variation trend of each concentration of oxygen and the supply variation trend of each interface within the same preset time period in the future. Based on the predicted trend of total oxygen demand over a future preset time period, the trend of total available oxygen at each concentration over the same future preset time period, and the trend of supply at each interface, the allocation time point when the total target oxygen demand exceeds the real-time available resource index is obtained.

8. The oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in claim 7, characterized in that, The process of obtaining the supply and demand allocation response strategy also includes: Based on the overall supply trend of each oxygen concentration and the supply trend of each interface after the allocation time point, combined with the overall demand trend of oxygen supply within the same time period, a multi-objective algorithm is used to solve the multi-objective oxygen supply allocation by combining the real-time oxygen supply priority index with the first optimization objective and the first optimization constraint, and the allocation solution is obtained. The first optimization constraint is: the real-time oxygen supply concentration and real-time oxygen supply flow rate corresponding to each oxygen supply target are greater than or equal to the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate corresponding to the current hierarchical mapping. The first optimization objective is: under the condition of satisfying the first optimization constraint, to make the supply satisfaction rate equal to 1 and the oxygen supply risk of all oxygen supply targets minimized. Based on the allocation solution results, when there is an allocation solution result that satisfies the supply satisfaction rate of 1, the standard supply oxygen concentration and the stratified oxygen supply standard range in the stratified mapping corresponding to each node in the oxygen demand graph after the allocation time point are adjusted according to the current solution result. The adjusted stratified mapping is used as the new supply and demand allocation response strategy until the total oxygen demand target is less than or equal to the real-time adjustable resource index, and then the supply and demand allocation response strategy corresponding to the original stratified mapping is restored.

9. The oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in claim 8, characterized in that, The process of obtaining the supply and demand allocation response strategy also includes: When there is no allocation solution that satisfies the supply satisfaction rate of 1, a forward search is performed using the allocation time point combined with the minimum safe standard oxygen supply concentration and minimum safe standard oxygen supply flow rate of each target to be supplied under the current hierarchical mapping until an allocation solution that satisfies the supply satisfaction rate of 1 is found, and the last time point of the forward search is taken as the new allocation time point. The hierarchical mapping after the new allocation time point is adjusted using the allocation solution results as a new supply and demand allocation response strategy until the total oxygen demand target is less than or equal to the real-time available resource index, at which point the supply and demand allocation response strategy corresponding to the original hierarchical mapping is restored.

10. An oxygen supply and demand regulation optimization system using a digital display oxygen flow meter, used to implement the oxygen supply and demand regulation optimization method using a digital display oxygen flow meter as described in any one of claims 1-9, characterized in that, include: The data acquisition module is used to acquire real-time oxygen demand indicators, real-time oxygen priority indicators, real-time adjustable resource indicators, real-time oxygen supply quality indicators, and abnormal operation indicators of the digital medical oxygen flow meter. The enhanced allocation module is used to initialize a preset oxygen supply and demand bidirectional mapping diagram. It combines real-time oxygen demand indicators, real-time oxygen priority indicators, and real-time adjustable resource indicators with a preset supply and demand bidirectional prediction model and an enhanced allocation control model to obtain supply and demand allocation response strategies. The simulation control module is used to respond to the supply and demand allocation response strategy on the oxygen supply and demand bidirectional mapping map within the preset oxygen supply simulation control platform, and to monitor real-time oxygen supply quality indicators and abnormal operation indicators of the digital medical oxygen flow meter. When either the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter fails to meet the preset constraint index value, the Peter-Clark algorithm is used to analyze and locate the real-time oxygen supply quality indicator or the abnormal operation indicator of the digital medical oxygen flow meter on the oxygen supply and demand bidirectional mapping map, and the location result is fed back to the enhanced allocation control model. Combined with the simulation algorithm, the supply and demand allocation response strategy is adjusted in real time, and the location result is simultaneously fed back to the configured display interface for real-time early warning display.

Images