Medical oxygen storage and distribution method and system based on internet of things

Abstract

This invention relates to the field of Internet of Things (IoT) sensing technology, and particularly to an IoT-based method and system for medical oxygen storage and distribution. The method includes the following steps: acquiring usage data of recovered medical oxygen storage containers; analyzing the pressure drop curves and usage frequency patterns of each container when oxygen is drawn using the usage data to establish usage behavior characteristics; obtaining container stability grading data by comparing the stability differences of the usage behavior characteristics of different containers; statistically analyzing the cumulative number of uses and load intensity changes of each container based on the container stability grading data; calculating the cumulative fatigue value of each container based on the cumulative number of uses and load intensity changes; and combining the cumulative fatigue value with the container stability grading data to evaluate container reliability and generate a container reliability ranking table. This invention, by analyzing the usage behavior characteristics and cumulative fatigue values ​​of containers, can identify containers with potential risks.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) sensing technology, and particularly to a method and system for medical oxygen storage and distribution based on the Internet of Things. Background Technology

[0002] Medical oxygen typically refers to oxygen with a purity of 99.5% or higher. A standard medical oxygen cylinder can withstand a pressure of 150 atmospheres, and a 40-liter cylinder, when fully pressurized, can store approximately 6 cubic meters of gaseous oxygen. The storage and transportation of medical oxygen requires special safety measures. Because oxygen is a strong oxidizer, it accelerates the combustion process and must therefore be kept away from sources of ignition and flammable materials. Medical oxygen is usually stored in high-pressure steel cylinders, which have special color markings and safety devices.

[0003] In the traditional model, industrial gas companies manage medical oxygen containers primarily through manual experience and simple record-keeping systems. The core difficulty lies in the lack of information transparency. Once a medical oxygen container is returned from a hospital, the industrial gas company knows almost nothing about its usage during that time. This information blind spot means that the container's condition assessment can only rely on visual inspections and basic pressure tests, which can easily lead to over-maintenance causing skyrocketing costs or under-maintenance resulting in safety risks. Summary of the Invention

[0004] Therefore, it is necessary for the present invention to provide a medical oxygen storage and distribution method and system based on the Internet of Things to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, an IoT-based medical oxygen storage and distribution method includes the following steps:

[0006] Step S1: Obtain usage data of the recovered medical oxygen storage containers; analyze the pressure drop curve and usage frequency pattern of each container when oxygen is drawn using the usage data to establish usage behavior characteristics; obtain container stability grading data by comparing the stability differences of usage behavior characteristics of different containers.

[0007] Step S2: Based on the container stability classification data, calculate the cumulative number of uses and load intensity changes of each container; calculate the cumulative fatigue value of each container based on the cumulative number of uses and load intensity changes; combine the cumulative fatigue value with the container stability classification data to evaluate the container reliability and generate a container reliability ranking table.

[0008] Step S3: Perform reliability matching on medical oxygen storage containers according to container reliability ranking, and divide them into reuse categories to obtain container reuse scheduling data. Specifically, the reuse categories are divided into high reliability containers as direct reuse category, medium reliability containers as reuse category after inspection, and low reliability containers as maintenance category or scrap category.

[0009] Step S4: Based on the container reuse scheduling data, perform classification processing and reallocation, quickly fill containers in the direct reuse category, verify the status of containers in the reuse category after testing, repair and restore the performance of containers in the maintenance category, and recycle the resources of containers in the scrap category, thereby realizing the reuse allocation of recycled medical oxygen storage containers.

[0010] The present invention also provides an Internet of Things (IoT)-based medical oxygen storage and distribution system for executing the above-described IoT-based medical oxygen storage and distribution method, wherein the IoT-based medical oxygen storage and distribution system comprises:

[0011] The behavior analysis module is used to acquire usage data of the recovered medical oxygen storage containers; it uses the usage data to analyze the pressure drop curve and usage frequency pattern of each container when oxygen is drawn, and establishes usage behavior characteristics; by comparing the stability differences of the usage behavior characteristics of different containers, it obtains container stability classification data.

[0012] The reliability assessment module is used to statistically analyze the cumulative number of uses and load intensity changes of each container based on the container stability classification data; calculate the cumulative fatigue value of each container based on the cumulative number of uses and load intensity changes; combine the cumulative fatigue value with the container stability classification data to assess the container reliability and generate a container reliability ranking table.

[0013] The intelligent scheduling module is used to perform reliability matching of medical oxygen storage containers according to the reliability ranking of containers, and to divide them into reuse categories to obtain container reuse scheduling data. Specifically, the reuse categories are marked as high reliability containers as direct reuse category, medium reliability containers as reuse category after inspection category, and low reliability containers as maintenance category or scrap category.

[0014] The classification execution module is used to perform classification processing and reallocation based on container reuse scheduling data. It performs rapid filling of containers in the direct reuse category, status verification of containers in the reuse category after testing, equipment repair and performance restoration of containers in the maintenance category, and resource recycling of containers in the scrap category, thereby realizing the reuse allocation of recycled medical oxygen storage containers.

[0015] This invention constructs a complete, efficient, and intelligent oxygen container recycling, assessment, and redistribution process, significantly improving the management efficiency, safety, and resource utilization of medical oxygen storage containers for industrial gas companies. This method introduces an IoT sensing system and a full-process data-driven mechanism into the recycling process, ensuring that the usage history of each container is accurately recorded and analyzed. In the acquisition phase, the container's unique identifier and its usage history data (including outbound time, usage location, sensor data, etc.) are collected synchronously and analyzed through time series analysis to construct a complete usage process chain. This identity association mechanism not only improves the traceability of container management but also provides a solid data foundation for subsequent status assessment. In the data processing phase, the system uses an algorithm model to perform integrity verification and cleaning of historical data, eliminating missing or abnormal records, and transforming the container's usage process into structured behavioral characteristics. Especially in pressure change behavior modeling, by analyzing the pressure drop curve of the container each time oxygen is used, key performance indicators such as the rate of drop, response delay, and holding capacity are captured. Simultaneously, combined with usage frequency patterns, usage regularity, intensity changes, and temporal distribution concentration are extracted, thereby establishing a behavioral characteristic matrix describing the state of each container. This behavioral characteristic model reflects the stability of containers in real medical environments, effectively revealing their fatigue risks and performance degradation trends. Furthermore, this method introduces a horizontal comparison mechanism at the behavioral characteristic level. This involves numerically calculating the differences in behavioral indicators across all containers within the same dimension to construct an outlier distribution matrix, quantifying the deviation of each container from the stable group, and classifying stability levels accordingly. This statistical analysis method eliminates traditional manual experience-based judgment, providing an objective, dynamic, and scalable container classification model. By assigning standardized identification codes to stability levels, a unified data foundation is laid for subsequent fusion calculations with other parameters. Based on this, the system further integrates container usage frequency and load intensity information to construct a cumulative usage frequency model and a historical pressure load model. Particularly in load intensity assessment, it not only collects the working pressure range during each use but also performs trend analysis and fluctuation identification on pressure fluctuations and peak usage pressure. Combining usage frequency and load intensity trends, a comprehensive fatigue cumulative value is calculated through a weighted combination method, forming a quantitative indicator of the service life consumption of each container. This fusion considers usage frequency, intensity fluctuations, overload behavior, and usage duration, comprehensively reflecting the container's service status and potential failure risks. At the comprehensive evaluation level, this method standardizes and weights the stability score and fatigue resistance score separately, and then integrates them, introducing a standard deviation ratio mechanism to determine the dynamics of their weight allocation, thereby forming a comprehensive reliability score for the container. To eliminate the influence of statistical bias, outlier identification and score verification mechanisms are also introduced to ensure the accuracy and robustness of the scoring model. This scoring system can be used for intuitive global ranking, providing a scientific basis for subsequent resource allocation.Based on reliability results, the method establishes a multi-level grading and mapping mechanism to transform scores into operable reuse categories. By statistically analyzing the overall score distribution of containers, the grading boundaries are calculated, and high-scoring containers are mapped to the direct reuse category, medium-scoring containers to the post-inspection reuse category, and low-scoring containers to the maintenance or scrap category. This grading and mapping rule not only has dynamic adaptability but also matches the actual supply chain status. At the resource scheduling level, this method introduces a supply and demand balance assessment algorithm to compare the container capacity distribution of each reuse category, identify potential supply and demand imbalances, and optimize the overall allocation structure by automatically adjusting the categories of some edge containers, ensuring business continuity. Finally, at the actual processing level, the method designs a parallel processing framework, placing containers of different reuse categories into rapid filling, inspection and verification, equipment maintenance, or resource recycling channels respectively. Each channel has independent process control logic and data acquisition mechanisms. Especially in the inspection and reuse queue, dynamic reclassification is performed based on container status verification results, and the processing plans for maintenance and scrapping channels are updated in real time, strengthening the closed-loop management of container status. The equipment maintenance process incorporates structural inspection, seal repair, and functional restoration operations to ensure recoverable containers meet standards again. The resource recycling process achieves harmless dismantling and material recovery, realizing environmental friendliness and material reuse goals. The entire method is data-driven at its core, supported by IoT devices for information sensing. Through the coordinated operation of multiple stages, including behavioral modeling, fatigue calculation, multi-dimensional assessment, classification scheduling, and parallel processing, it constructs an intelligent recycling and reuse solution covering the entire lifecycle, all-state determination, and all-process handling of containers. This method not only significantly improves the recycling and reuse efficiency of medical oxygen containers, reduces resource waste and potential safety hazards, but also achieves fully transparent management and precise redistribution of container status. It is suitable for the construction of container recycling systems in large-scale industrial gas companies and is an important technical support path for promoting the transformation and upgrading of smart medical gas logistics management. Attached Figure Description

[0016] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0017] Figure 1 This is a schematic diagram of the steps of a medical oxygen storage and distribution method based on the Internet of Things according to the present invention.

[0018] Figure 2 for Figure 1 A detailed flowchart of step S1;

[0019] Figure 3 This is a schematic diagram of the portable medical oxygen supply device of the present invention;

[0020] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0022] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0023] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0024] To achieve the above objectives, please refer to Figures 1 to 3 This invention provides a method for storing and distributing medical oxygen based on the Internet of Things (IoT), the method comprising the following steps:

[0025] Step S1: Obtain usage data of the recovered medical oxygen storage containers; analyze the pressure drop curve and usage frequency pattern of each container when oxygen is drawn using the usage data to establish usage behavior characteristics; obtain container stability grading data by comparing the stability differences of usage behavior characteristics of different containers.

[0026] This invention deploys medical oxygen storage containers with embedded sensors in various medical institutions. Each container is equipped with a uniquely identified RFID chip, a pressure sensor, a temperature sensor, and a position sensor, and transmits data in real time to the company's backend server via an NB-IoT network. When a container is used by a hospital, the pressure sensor collects the pressure data of the oxygen output in real time, recording the pressure value once per second, and uploads the full-cycle pressure change data after the container is recycled. The platform decodes this data, extracts typical pressure drop segments when oxygen is used, and calculates the start time, pressure drop rate, total drop magnitude, and duration of each drop event, forming a "pressure drop curve set." Simultaneously, based on continuous logs of container outbound and recycled times, the platform statistically analyzes the daily, weekly, and monthly usage frequency and usage time periods of each container, identifying frequency patterns such as whether it is used daily or concentrated in specific time periods, generating "usage frequency pattern data." Subsequently, machine learning clustering algorithms (such as DBSCAN) are used to perform group analysis on the usage behavior characteristics of all containers, identify container groups with highly consistent usage behavior parameters, and regard containers with large deviations as individuals with poor stability. Thus, container stability is divided into three categories: A (high stability), B (medium stability), and C (low stability), and finally, "container stability classification data" is generated and uploaded to the enterprise's internal container health database.

[0027] Step S2: Based on the container stability classification data, calculate the cumulative number of uses and load intensity changes of each container; calculate the cumulative fatigue value of each container based on the cumulative number of uses and load intensity changes; combine the cumulative fatigue value with the container stability classification data to evaluate the container reliability and generate a container reliability ranking table.

[0028] Based on the obtained container stability grading data, this invention retrieves the complete lifecycle records of all containers from their initial use to the present. This data is automatically traced through an IoT platform, tracking the outbound and return logs of each container and archiving them by container number to form an "individual usage file." The system calculates the total number of outbound trips for each container, records the oxygen filling time, outbound pressure, minimum pressure during use, and remaining pressure upon return during each lifecycle, and calculates the average working pressure and pressure variation for each use. Then, based on a timeline, the above usage records are integrated to obtain the container's "cumulative usage count" and "load intensity variation trajectory." Next, the platform calculates the container utilization rate according to the industry-set design lifespan (e.g., 800 outbound uses). A utilization rate exceeding 640 times (i.e., 80%) is considered high and assigned a high risk factor. Simultaneously, the platform calculates the ratio of the container's average working pressure to the container's nameplate design standard pressure (e.g., 15 MPa) to determine if overload behavior exists. If the ratio exceeds 1.2, it is marked as overloaded. By combining the trend of load changes (e.g., a continuous increase in recent usage pressure) and the fluctuation range (e.g., pressure fluctuations exceeding ±20% each time), trend adjustment coefficients and fluctuation penalty coefficients are assigned respectively. Finally, the platform comprehensively considers three types of factors: utilization rate risk, overload coefficient, and pressure change coefficient, and uses a weighted method to score the fatigue strength of the containers. For example, the impact of utilization rate is weighted at 60%, pressure load at 25%, and pressure fluctuation at 15%. From this, the "comprehensive fatigue cumulative index" of each container is calculated, and combined with the stability level, a "container reliability score" is generated through a two-factor mapping. The containers are then sorted from high to low scores to form a "container reliability ranking table".

[0029] Step S3: Perform reliability matching on medical oxygen storage containers according to container reliability ranking, and divide them into reuse categories to obtain container reuse scheduling data. Specifically, the reuse categories are divided into high reliability containers as direct reuse category, medium reliability containers as reuse category after inspection, and low reliability containers as maintenance category or scrap category.

[0030] In this embodiment of the invention, after obtaining the container reliability ranking table, the system retrieves the current total inventory of containers and the real-time oxygen demand forecast data of the client's medical institution. An algorithm is used to group and filter the reliability ranking, marking containers with the top 20% reliability scores as "high reliability" and directly including them in the "direct reuse category"; containers with scores between 20% and 70% are classified as "medium reliability" and assigned to the "reuse category after inspection"; and containers with scores in the bottom 30% are identified as "low reliability" and marked as "maintenance or scrapping category". This grouping is dynamically adjusted based on the mean and standard deviation of the reliability distribution to ensure that the allocation ratio adapts to the current supply and demand situation. The system automatically establishes a mapping table between container numbers and reuse categories and transmits it to the scheduling engine. Simultaneously, it counts the number of containers in each category and, combined with the hospital's monthly peak and valley gas demand, predicts the demand gap for each category of containers in the coming week. If the number of medium category containers is insufficient, the system dynamically promotes some low-level containers close to the upper-middle score boundary to alleviate the gap, forming the final "container reuse scheduling data". This data includes fields such as container number, reliability level, processing category, and recommended scheduling priority, providing a foundation for subsequent automated processing.

[0031] Step S4: Based on the container reuse scheduling data, perform classification processing and reallocation, quickly fill containers in the direct reuse category, verify the status of containers in the reuse category after testing, repair and restore the performance of containers in the maintenance category, and recycle the resources of containers in the scrap category, thereby realizing the reuse allocation of recycled medical oxygen storage containers.

[0032] Upon receiving container reuse scheduling data, this embodiment of the invention automatically initiates a "classification and processing flow," constructing four types of physical processing queues at the enterprise's packaging plant: "Direct Filling Area," "Status Verification Area," "Repair Area," and "Resource Recycling Area." The system automatically transports each container to its corresponding workstation according to the container number in the scheduling data. On the "Direct Reuse Category" processing line, containers undergo automatic pressure testing before entering a rapid filling process. Each container is filled to 14.8 MPa and automatically completes pressure relief and sealing tests, with the entire process completed within 12 minutes. In the "Detection" process… In the "Reuse Category" processing line, the platform retrieves historical usage records and sensor residual pressure data, and activates an AI image-assisted detection system to check for issues such as weld aging and valve corrosion. Containers that pass verification are remarked as "filling and ready for use," while those that fail are automatically transferred to the repair or recycling area. In the "Maintenance Category" processing line, the system automatically performs internal wall cleaning, parts replacement, and airtightness re-inspection. After repair, maintenance records are uploaded to the container's electronic file. For containers in the "Scrap Category," they are uniformly guided to the resource recycling area for depressurization, metal sorting and dismantling, and material recycling. All processing data is automatically uploaded to the execution monitoring platform, generating a "Processing Execution Report" containing the container number, processing path, operation time, and anomaly handling. This report serves as part of the container's full lifecycle management and supports the next round of scheduling analysis and operational optimization.

[0033] Preferably, step S1 includes the following steps:

[0034] Step S11: Synchronously read the identification code of the recycled medical oxygen storage container and the stored usage history data through the Internet of Things sensor network, wherein the usage history data includes the time of release, the location of use, the time of recycling, and the stored sensor data;

[0035] In this embodiment of the invention, after the medical oxygen storage container is recycled, the data reading program for the recycled container is immediately initiated through IoT gateway nodes deployed at logistics transfer stations and factory entrances. The IoT sensor network automatically identifies the RFID or NFC electronic tags bound to the container, obtains the container's unique identification code, and upon successful identification, synchronously wakes up the container's embedded controller, retrieving complete usage history data from its internal Flash storage module. This historical data includes the outbound time (i.e., the time the container is sent from the enterprise's warehousing system to the hospital), the usage location (recorded by the GPS positioning module and uploaded via the wireless communication module), the recycling time (the system's receiving timestamp when the container is returned by the hospital), and the raw sensor data collected by embedded pressure sensors, temperature and humidity sensors, etc., throughout the entire usage cycle. All data is preprocessed and encrypted by edge computing devices before being uploaded to the enterprise's cloud platform's container information control system. During the upload process, lightweight transmission is ensured through the MQTT protocol to guarantee transmission stability. Finally, a "raw data entry" is generated for each container on the platform.

[0036] Step S12: Arrange the usage history data and identification code in chronological order to form a continuous usage process record, and obtain container identity association data;

[0037] After successfully receiving the original data entry in step S11, this embodiment of the invention automatically appends the container's historical data to the container's master data table in the platform database based on the container's identifier code. It then calls a data sorting program to sort all the container's usage history by time, ensuring that nodes such as outbound time, usage process, and recycling time form a continuous time-series data chain, generating a "usage process record sequence." During this process, the platform establishes a "container identity-associated data object" on a container-by-container basis. This object uses the identifier code as the key value, and its internal structure includes continuous outbound-usage-recycling records, a timestamp sequence of sensor data, boundary markers for each usage cycle, and the location code of the associated hospital or user unit. This type of "container identity-associated data" is the core structure for container status tracking. After completing the construction of this data object, the platform automatically writes the status flag "archived," preparing the structure for subsequent data cleaning and analysis.

[0038] Step S13: Perform integrity verification and data cleaning on the container identity association data, check for missing data, incorrect timestamps or sensor malfunction records, and mark the time periods with incomplete data to obtain usage history data;

[0039] In this embodiment of the invention, after the container identity association data is constructed, the platform automatically triggers the integrity verification module to perform quality checks on each record. First, it checks whether key fields contain null values ​​or abnormal characters, such as whether the outbound time is later than the return time, whether there are breakpoints in the sensor pressure values ​​(continuous missing values ​​exceeding 10 minutes), and whether the timestamps are in reverse order. If any missing or abnormal data is found, the corresponding time period is marked as an "incomplete data segment," and a corresponding error type label is generated, such as "time logic error," "sensor interruption," or "data drift." Subsequently, the system performs data cleaning, using time series interpolation algorithms to estimate and complete missing values, correcting erroneous timestamps based on regression inference from adjacent data, and replacing or removing obvious outliers with median values. Furthermore, for data anomalies caused by sensor malfunctions (such as pressure consistently at 0 or pressure suddenly increasing to extremely high values), the system compares the duration of the anomaly with historical behavior patterns to determine if it is a hardware failure. If confirmed, the data segment is marked as unusable and recorded in the container anomaly log. Finally, the system outputs "Historical data cleaned and used," updates the container status to "analyzable," and archives the abnormal segments for manual verification.

[0040] Step S14: Analyze the pressure drop curves and usage frequency patterns of each container when oxygen is drawn using the usage process data to establish usage behavior characteristics;

[0041] After obtaining the cleaned usage history data, the system of this invention enters the behavior modeling process. First, it analyzes the sensor pressure time series of each container and uses a sliding window mechanism to identify "pressure drop events," which are defined as a continuous pressure drop within a unit of time that meets a preset threshold change (e.g., a drop exceeding 0.5 MPa and lasting for more than 30 seconds). The platform records the start time, end time, drop rate (drop amount divided by time), minimum pressure value, and recovery speed of this event. This information forms a "pressure drop curve." At the same time, the platform statistically analyzes the past 1 month, 3 months, and 6 months for each container. Parameters such as usage frequency, usage time period distribution, and number of uses per unit time are used to determine whether there is periodic, concentrated, or intermittent usage behavior. These constitute the "usage frequency pattern." For example, if some containers are used frequently in the morning and almost never in the evening, it indicates that there is a fixed usage pattern. The platform uses the K-Medoids clustering method to aggregate and analyze the behavioral characteristics and generate a "usage behavior feature vector" for each container. The vector includes 6 to 8 quantitative indicators in dimensions such as pressure drop rate volatility, usage frequency stability, and pressure response consistency, which serve as a behavioral profile and lay the feature foundation for subsequent stability analysis.

[0042] Step S15: By comparing the stability differences in the usage behavior characteristics of different containers, container stability grading data is obtained.

[0043] In this embodiment of the invention, after the usage behavior feature vectors of all containers are generated, the system enters the stability assessment process. First, the standard deviation and mean of each feature dimension in the container group are calculated, and a "behavioral baseline curve" is established with highly stable containers as a reference. Then, the Euclidean distance between each container's behavior feature vector and the baseline curve is calculated. This distance measures the degree of deviation from the stable group; the smaller the deviation, the more stable the container. The platform constructs an "outlier distribution matrix" from the deviation values ​​of all containers and sets stability boundary points (e.g., 1.0 and 2.0 standard deviations as medium-to-high boundaries) using a distribution fitting method (such as Gaussian distribution fitting). This classifies container stability into three levels: high stability, medium stability, and low stability, and assigns labels A, B, and C respectively. Finally, the platform outputs a "container stability grading data table," where each row corresponds to a container identifier and its corresponding stability level and quantitative score. This data is used in subsequent fatigue assessment and scheduling allocation stages, and the stability score will be continuously updated as one of the core indicators for container quality assessment.

[0044] Preferably, step S14 includes the following steps:

[0045] Step S141: Extract the pressure change time series from the historical data, identify the pressure drop events corresponding to the customer's oxygen intake operation, and record the starting point, rate of decrease and ending point of each pressure drop by continuously monitoring the downward trend and magnitude of the pressure value to obtain the pressure drop curve.

[0046] After completing the preliminary data cleaning, the IoT platform deployed in this embodiment of the invention reconstructs the sensor data of each container into a pressure change time series along a time axis, typically with a sampling frequency of 10 seconds per sampling. The platform system uses a continuous sliding window method to segment and detect the pressure data, determining whether a continuous downward trend exists within each time period. A threshold is set as a continuous pressure drop of more than 0.2 MPa for a duration of not less than 90 seconds, which is considered a "pressure drop event". For each event, the platform automatically records parameters such as the event start time (i.e., the time when the pressure first significantly drops), the event end time (i.e., the time when the pressure tends to stabilize or slightly recover), the rate of decrease (total decrease divided by duration), and the lowest pressure value. These events are then organized and numbered on the time axis, and the pressure change process is fitted as a curve to generate a "pressure drop curve". This curve represents the actual gas consumption process when the client takes medical oxygen from the container, and is used for subsequent extraction of behavioral patterns and response characteristic analysis.

[0047] Step S142: Based on the usage process data, statistically analyze the usage frequency and time distribution characteristics of each container, calculate the average number of uses per unit time, the regularity of the usage time interval, and the periodic changes in usage intensity to obtain the usage process time characteristics;

[0048] Based on the aforementioned extracted set of pressure drop events, this invention's embodiments utilize a platform system to perform time distribution statistical analysis on the historical usage cycles of each container, focusing on pattern recognition of usage frequency and intervals within a unit of time. For example, by dividing time periods into weekly units, the system statistically analyzes indicators such as the number of times the container is used each week, the daily usage time distribution, and the average daily usage duration. Further analysis is conducted to determine if the usage intervals exhibit regularity, such as whether it is used at a fixed interval of 8 hours, or whether oxygen is frequently drawn only during daytime or working hours. The platform also identifies specific periodic characteristics, such as whether the container is only used during certain periods or whether there are instances of high-intensity, short-duration concentrated use. The statistical results include the average number of weekly uses, typical usage intervals (e.g., 6-hour intervals), and the distribution range of usage duration, forming "time characteristics of the usage process," used to characterize the container's workload and usage patterns in a real medical environment.

[0049] Step S143: Generate the usage frequency pattern of each container by analyzing the concentration and dispersion of the usage process time characteristics;

[0050] After obtaining the "time characteristics of the usage process," this embodiment of the invention further uses data mining methods to analyze its concentration and dispersion. Concentration reflects whether the usage frequency is concentrated in a certain time period, such as high frequency concentrated in the daytime from 9:00 to 11:00. Dispersion describes whether the usage time point is spread throughout different times of the day. The system calculates the kurtosis and skewness of the distribution by histogram analysis of the time distribution, and determines the concentration or dispersion characteristics in time by combining the coefficient of variation, information entropy and other means. At the same time, it classifies and statistically analyzes different time windows (such as weekdays vs. weekends) to assess whether there are rhythmic differences. The platform finally generates the "usage frequency pattern" for each container, classifies it into labels such as high regularity, periodicity, dispersion, and suddenness, and generates structured parameters including usage density index and time concentration index as the basis for subsequent feature modeling.

[0051] Step S144: Extract features and quantify the pressure drop curve and usage frequency pattern. Extract the stability of the drop rate, the consistency of the response time, and the pressure holding capability from the pressure drop curve. Extract the frequency regularity, the adaptability of usage intensity, and the uniformity of time distribution from the usage frequency pattern, thereby establishing usage behavior characteristics that describe the performance of the container.

[0052] In this embodiment of the invention, after constructing the pressure drop curve and usage frequency pattern, the industrial gas company's IoT platform initiates a behavioral feature extraction module to perform quantitative analysis on two dimensions. For the pressure drop curve, key extracted features include the stability of the drop rate (i.e., the standard deviation of the rate across multiple drop events, reflecting the consistency of the gas supply valve's response), the consistency of response time (the mean and deviation of the duration of different drop events), and the pressure holding capacity (whether the stable section after pressure drop remains within the safe output range, for example, between 0.4 and 0.6 MPa). For the usage frequency pattern, the platform extracts frequency regularity (e.g., whether there are periodic usage peaks), usage intensity adaptability (e.g., whether it can withstand high-frequency operation in a short period), and time distribution uniformity (whether it avoids concentration in a certain period). The platform normalizes the quantitative parameters of these multiple dimensions and generates a "usage behavior feature vector" for each container. This vector serves as a digital label describing the container's stability, pressure resistance, and adaptability in actual use, and is directly input into the subsequent stability modeling and reliability assessment processes.

[0053] Preferably, step S15 includes the following steps:

[0054] Step S151: Perform a horizontal comparison of the usage behavior characteristics of all recycling containers, calculate the numerical differences and distribution range of the usage behavior characteristic parameters of each container, and obtain the outlier distribution matrix data of container usage behavior.

[0055] In the preliminary steps of this invention, the industrial gas company has generated usage behavior feature vectors for each recovered medical oxygen storage container through an IoT platform. These vectors include multiple dimensions, such as descent rate stability, response time consistency, pressure retention capacity, usage intensity adaptability, and time distribution uniformity. In this step, the platform system performs a horizontal comparative analysis of the behavior feature vectors of all recovered containers. This involves inputting the feature vector set of all containers as a multi-dimensional feature matrix, eliminating dimensional differences through standardization (e.g., min-max normalization or Z-score standardization), and then using principal component analysis (PCA) or feature distribution aggregation algorithms to identify the numerical distribution range and overall variance of each dimension. The platform further uses box plot analysis and Local Outlier Factor (LOF) algorithms to detect data points that significantly deviate from other containers in each feature dimension, forming multi-dimensional "outlier scores." These outlier scores are then aggregated by container number to form a "container usage behavior outlier distribution matrix data." This matrix uses containers as rows and the degree of outlier in each behavioral feature dimension as columns, comprehensively reflecting the degree of deviation of the container's usage behavior relative to the group, providing fundamental data support for subsequent stability assessments.

[0056] Step S152: Identify stable and unstable container groups based on the outlier distribution matrix data, quantify the deviation of each container from the benchmark of the stable container group, and obtain the stability difference;

[0057] In this embodiment of the invention, after obtaining the outlier distribution matrix data of container usage behavior, a baseline container group identification module is activated. This module automatically clusters the outlier vectors of all containers using clustering analysis algorithms (such as K-means or DBSCAN), classifying containers with overall low outlier scores as a "stable group." This group is representative, exhibiting small fluctuations in behavioral characteristics, concentrated parameters across dimensions, and low deviation values. The platform uses the characteristic mean of this stable group as a "behavioral characteristic stability benchmark" and calculates the Euclidean or Mahalanobis distance from this benchmark for all other containers to quantify the degree of deviation of their behavioral performance from stability. The platform system defines this distance value as a "stability difference score," with a larger value indicating that the container's behavior pattern deviates more from the overall stable reference model. Finally, the score result is stored in the database as the core basis for subsequent hierarchical modeling and forms a visualized stability distribution map, used by the enterprise decision-making system to identify core container resource groups with good operational stability and abnormal containers.

[0058] Step S153: Establish a container stability grading standard based on the numerical distribution of stability differences, and assign a corresponding identification code to each grade to obtain container stability grading data.

[0059] Based on the stability difference scores obtained in the previous step, the industrial gas company establishes a grading model module in this embodiment of the invention. The platform first performs statistical analysis on the distribution of stability difference scores for all containers, evaluating characteristic parameters such as mean, median, standard deviation, and distribution pattern. According to the distribution results, the system sets stability grading standards according to statistical intervals: for example, containers with stability difference scores within ±0.5 times the group average standard deviation are classified as "Class I stable containers," those within ±1.0 times are "Class II moderately stable containers," and those exceeding 1.0 times are "Class III unstable containers." Containers with extremely outlier scores (e.g., exceeding 2 times the standard deviation of the mean) are marked as "abnormal" containers. The system assigns a unique identifier code (e.g., A, B, C, D) to each stability level and automatically matches each container to the corresponding level, ultimately generating "container stability grading data." This data includes fields such as container number, stability score, level identifier, and basic characteristic summary, serving as important sub-indicators for subsequent reliability assessments, ensuring that the industrial gas company can implement refined container scheduling and resource management based on real-world stability data.

[0060] Preferably, step S2, which involves statistically analyzing the cumulative number of uses and load intensity changes of each container based on container stability grading data, includes:

[0061] Based on the container identification information in the container stability classification data, extract the outbound and return records of each container since it was put into use, and classify and organize them according to the container number to form an individual usage file for each container.

[0062] Based on the total number of times each container is dispatched according to individual usage records, a container dispatch statistics table is obtained.

[0063] Based on the container outbound frequency statistics table, calculate the distribution of usage frequency of each container in different time periods, and perform cumulative summation. Add the usage frequency of each time period step by step in chronological order to calculate the cumulative usage frequency from the time of use to the current time, and obtain the cumulative usage frequency of each container.

[0064] Extract pressure load information for each use process from individual usage records, including the filling pressure value at each time of departure, the working pressure range during use, and the remaining pressure value at the time of recovery. Calculate the average working pressure, maximum working pressure, and pressure change range of each container in each use to obtain historical load intensity data of the container.

[0065] By comparing the differences in average working pressure over different periods based on historical load intensity data of the container, the trend of load intensity can be identified, and the change in load intensity can be obtained.

[0066] This invention maintains electronic identification codes (such as RFID codes or QR code serial numbers) for all medical oxygen storage containers on an IoT platform. These codes are used throughout the entire process of container filling, transportation, use, and recycling. The platform uses the container identification information from the "container stability grading data" generated in the preceding steps as the primary key field, automatically calling the logistics record table and sensor data table in the data warehouse with fields such as "outbound time," "recycling time," "filling pressure," and "customer ID." The platform uses an automatic SQL retrieval function to aggregate and sort the outbound and recycling activities of each container by its number, and arranges each group of records along a timeline to form an "individual usage file." This file not only includes the time point of each container's deployment, the customer served, the oxygen type, and the usage time, but also integrates sensor data such as pressure and temperature collected during use, serving as the foundational data structure for subsequent statistical analysis. Aggregation and statistical operations are performed on the constructed individual usage files, traversing and recording the number of outbound records from the first deployment to the current time by container number. Considering that some containers may be out of service due to maintenance or testing, the platform sets filtering conditions during statistics, only counting outbound records marked as "normal delivery". For example, container number OX-003257 has recorded 68 outbounds since its launch in 2021, of which 59 were official uses, and the rest were internal transfers or testing operations. The statistics system integrates the official outbound counts of all containers into a "Container Outbound Count Statistics Table". This table uses the container number as the index field and the outbound count as the statistical field, serving as a quantitative indicator of usage frequency to support fatigue level calculations. To more accurately analyze the usage rhythm and intensity fluctuations of containers, the platform performs time-slicing processing on a monthly basis for each container's outbound behavior, classifying each outbound date by month, for example, 3 outbounds in March 2022, 2 outbounds in April 2022, and so on. The platform cumulatively adds up the outbound counts for each month in chronological order to obtain the "cumulative usage count time series" for each container. For container OX-003257, as of June 2025, its cumulative outbound usage was 59 times, which is the "cumulative usage count" at the current point in time. This time series is not only used for static statistics in this step, but also lays the foundation for subsequent identification of load trends and assessment of fatigue changes. Pressure sensor data for each usage cycle is retrieved from the individual usage record; the system's preset sensor recording frequency is once every 30 seconds. The platform automatically identifies the "initial filling pressure" (i.e., the maximum pressure value measured before outbound), the "working pressure range" (taking the middle 80% range of pressure values ​​during customer use as the stable value range), and the "remaining pressure" (the pressure measured on the day of recovery) for each cycle. For example, in a certain usage record, the filling pressure of container OX-003257 was 14 MPa, the stable working range was 9 to 11 MPa, and the recovery pressure was 6 MPa. The platform calculates the pressure difference and working range fluctuation range for each usage cycle and records the maximum working pressure.After averaging multiple usage records, the container's historical "average working pressure," "maximum working pressure," and "pressure variation per unit usage cycle" are obtained, collectively referred to as "container historical load intensity data." The container's historical load intensity data is aggregated by time period (e.g., quarterly or semi-annual), the average working pressure for each period is calculated, and the numerical changes between adjacent time periods are compared. If the container shows a significant upward trend in the past two years, for example, the average working pressure increases from 9.3 MPa in Q1 2023 to 11.2 MPa in Q2 2025, the platform marks it as a "load intensity increasing trend." The platform sets a judgment threshold; for example, if the increase exceeds 10% for two consecutive time periods, it is considered a "significant increase." Simultaneously, if the pressure fluctuation is too high (standard deviation exceeds 0.8 MPa) in any time period, the platform records it as "significant load fluctuation." These indicators, when summarized, constitute the container's "load intensity change" data, serving as one of the input parameters for fatigue assessment. Combined with outbound frequency and pressure level, this data identifies potential high-stress aging risks in the container.

[0067] Preferably, step S2, which calculates the cumulative fatigue value of each container based on the cumulative number of uses and changes in load intensity, includes:

[0068] The utilization rate percentage is calculated by dividing the cumulative number of times each container is used by its designed service life, thus obtaining the container utilization rate data.

[0069] Risk classification is performed based on container usage data. When the usage data exceeds 80%, it is marked as high usage and assigned a risk coefficient of 1.5. When the usage data is below 80%, it is marked as standard usage and assigned a risk coefficient of 1.0, thus obtaining the usage risk coefficient.

[0070] The load intensity ratio data is obtained by calculating the ratio of the average working pressure to the design standard pressure based on the historical load intensity data of the container.

[0071] Overload risk assessment is performed based on the load intensity ratio. When the load intensity ratio exceeds 1.2, it is marked as overloaded use and a penalty coefficient of 2.0 is assigned. When the load intensity ratio is below 1.2, it is marked as normal use and a penalty coefficient of 1.0 is assigned. The overload coefficient is then obtained.

[0072] Based on the analysis of load intensity changes, load change trends and fluctuation characteristics are determined. When the load shows an upward trend, a trend coefficient of 1.3 is assigned, and when the load fluctuation exceeds 50% of the average value, a fluctuation coefficient of 1.4 is assigned, thus obtaining the load change adjustment coefficient.

[0073] The comprehensive fatigue calculation is based on the utilization rate risk coefficient, overload coefficient and load change adjustment coefficient. The container utilization rate data is multiplied by the utilization rate risk coefficient and then by 60% weight, plus the load intensity ratio multiplied by the overload coefficient and then by 25% weight, plus the load change adjustment coefficient multiplied by 15% weight, to obtain the comprehensive fatigue cumulative index.

[0074] Risk classification is performed based on the comprehensive fatigue accumulation index to obtain the container fatigue accumulation value.

[0075] In this embodiment of the invention, the design lifespan of each medical oxygen storage container is preset during the container production and procurement stage. For example, the design lifespan of a standard steel cylinder is 800 uses. Based on the "cumulative usage counts" collected in previous steps, the platform performs a calculation of "usage counts ÷ design lifespan counts × 100%" for each container. For example, container number OX-003257 has been used 640 times, and its usage rate is 80%. After the system completes the calculation for all containers in batches, it generates a "Container Usage Rate Data Table," where each container ID corresponds to a usage rate percentage field, used to measure the current consumption level of the container during its lifecycle. Based on the "Container Usage Rate Data Table," threshold grading rules are set to automatically determine whether the container usage rate exceeds 80%. Containers exceeding this threshold are marked as "high usage rate" and a value of 1.5 is written in the corresponding field; containers below the threshold are marked as "standard usage rate" and a value of 1.0 is written in the corresponding field. For example, container OX-003257 has a utilization rate of exactly 80%, which falls into the high utilization rate category according to business rules. The system automatically records 1.5 in its "Utilization Rate Risk Coefficient" column, indicating that this container requires a higher risk amplification factor in fatigue assessment. All results are automatically generated into a utilization rate risk coefficient table for subsequent fatigue index weighted calculation. The standard design pressure of the industrial gas company is 12 MPa. In previous steps, the average working pressure of each container during its use has been calculated. The platform calculates the "load intensity ratio" based on the method of "average working pressure ÷ standard design pressure". For example, if the average working pressure of container OX-003257 is 13 MPa, then its load intensity ratio is 1.08. The system writes the results of all containers into the "Load Intensity Ratio Table". This data reflects whether the container is at risk of operating under high pressure for a long time, which in turn affects the rate of material fatigue accumulation. The "Load Intensity Ratio Table" is imported into the risk assessment model, and logical judgments are automatically executed based on the threshold of 1.2. If the ratio of the container is higher than 1.2, it indicates that it is operating beyond its design load, and the system marks it as "overloaded use" and records a penalty coefficient of 2.0. If the ratio is less than or equal to 1.2, it is recorded as "normal use" with a coefficient of 1.0. For example, the load ratio of container number OX-004112 is 1.25, and the platform automatically assigns a penalty coefficient of 2.0, indicating that a more serious structural fatigue risk needs to be considered in the assessment. Using the average working pressure sequence of each quarter statistically analyzed in the previous step, it is determined whether there is a gradual upward trend. If at least three of the past four time periods show an increase in pressure, the system determines it as a "load upward trend," and the trend coefficient is set to 1.3. At the same time, the system calculates the ratio of the standard deviation of each pressure value to the overall average value. If the ratio exceeds 50%, it is considered "significant fluctuation," and the system assigns a fluctuation coefficient of 1.4.The platform ultimately combines the trend coefficient and the fluctuation coefficient to form the "load change adjustment coefficient." If only the trend condition is met, the coefficient is 1.3; if both are met, the coefficient is 1.82 (i.e., 1.3 multiplied by 1.4), and this is recorded in the fatigue analysis file of the corresponding container. All intermediate result tables are imported into the fatigue index calculation module, which automatically performs weighted summation of the three indicators. For container number OX-003257, its utilization rate is 80%, risk coefficient is 1.5, load ratio is 1.08, overload coefficient is 1.0, and load adjustment coefficient is 1.3. The system multiplies its utilization rate by the risk coefficient (resulting in 120), then multiplies it by a weight of 0.6 to obtain the main contribution value of 72; then multiplies the load ratio by the penalty coefficient (resulting in 1.08), then multiplies it by 0.25 to obtain 27%; finally, it multiplies the load change coefficient of 1.3 by 0.15 to obtain 19.5%. The final summation of the three factors yields the comprehensive fatigue cumulative index: 72 + 27 + 19.5 = 118.5 (the unit can be fatigue integral). After batch processing of all containers, a "Fatigue Cumulative Index Table" is generated. Fatigue grading standards are set based on an empirical model: a fatigue index less than 90 is considered "mild fatigue," 90 to 130 is "moderate fatigue," and over 130 is "severe fatigue." The platform matches the fatigue index of each container with the corresponding standard range, labels it with its fatigue level, and outputs it to the "Container Fatigue Cumulative Value Table." For example, container number OX-003257 has an index of 118.5, placing it at the moderate fatigue level, with the corresponding fatigue level code "F2." This result will be used for subsequent reliability ranking and reuse scheduling, ensuring that high-risk containers are no longer directly filled and used clinically, thus constructing a full lifecycle management model for industrial gas companies with high safety requirements.

[0076] Preferably, step S2, which combines the fatigue accumulation value with the container stability grading data to assess the container reliability, includes:

[0077] Based on the stability index values ​​in the container stability grading data, the stability index values ​​are multiplied by a standardization coefficient and then subjected to interval mapping to transform the value range of the stability index to a scoring range of 0-100, thus obtaining the container stability score.

[0078] The container fatigue resistance score is obtained by performing reverse numerical transformation based on the cumulative fatigue value of the container and multiplying the reciprocal of the cumulative fatigue value by a preset standardization factor.

[0079] The relative weights are determined by calculating the ratio of the standard deviations of the two types of scores, namely, the container stability score and the fatigue resistance score, and the dynamic weight allocation parameters are obtained.

[0080] The container stability score and fatigue resistance score are weighted and averaged according to the dynamic weight allocation parameters to obtain a preliminary comprehensive reliability score.

[0081] Based on the preliminary comprehensive reliability score, the standardized deviation of each container score from the group average is calculated. When the absolute value of the standardized deviation exceeds twice the standard deviation, it is marked as a statistical outlier. The outliers are then verified by data tracing and recalculated to obtain the verified comprehensive reliability score.

[0082] The reliability levels are determined based on the comprehensive reliability score after verification, resulting in a container reliability ranking table.

[0083] In this embodiment of the invention, a set of stability index values ​​is calculated for each container during the container stability grading stage. These values ​​include behavioral characteristic stability and pressure response stability, and their original values ​​may be distributed across different dimensions. The platform performs a linear range transformation on each index by setting a unified standardization coefficient (e.g., mapping the maximum value to 100 and the minimum value to 0). The system then weights and integrates the transformation results to obtain a container stability score in the range of 0 to 100. For example, the original stability value of container number OX-005148 is 0.86, and its stability score after mapping is 86.2. This score is used to measure the consistency and predictability of the container's performance in actual use. The fatigue cumulative value of each container has been obtained in the previous stage; the larger the value, the more severe the structural fatigue. To obtain a positively represented fatigue resistance, the platform takes the reciprocal of each fatigue cumulative value, so that containers with smaller values ​​receive higher scores. Combined with a set standardization factor (e.g., setting the maximum resistance score to 100), the container fatigue resistance score transformation is completed. Taking container OX-005148 as an example, its cumulative fatigue value is 112.5, with a reciprocal of approximately 0.00889. After standardization and multiplication by 1130, the fatigue resistance score is approximately 100.6, and the final value is limited to 100. This score directly reflects the container's ability to resist structural losses and, together with the stability score, constitutes the subsequent reliability index. The standard deviations of the stability and fatigue resistance scores for all containers are calculated within the overall container sample to determine the degree of variation of the two scoring dimensions in the current sample distribution. The platform automatically derives dynamic weighting parameters by calculating the ratio of the two standard deviations. For example, if the standard deviation of the stability score is 12 and the standard deviation of the fatigue resistance score is 24, the system will consider the fatigue resistance difference to be greater and assign it a higher weight. The platform further converts the ratio into weighting parameters, such as setting the stability score weight to 33% and the fatigue resistance score weight to 67%, for the next step of comprehensive scoring. Based on the aforementioned two scores and corresponding weights, a weighted average method is used to obtain the preliminary comprehensive reliability score for each container. For example, container OX-005148 has a stability score of 86.2 and a fatigue resistance score of 100. In the current batch, the weights allocated to stability score are 30% and fatigue resistance score is 70%. The system automatically performs a weighted summation of "score × weight," resulting in a preliminary comprehensive score of: 86.2 × 0.3 + 100 × 0.7 = 95.86. All containers generate an unverified preliminary score dataset using this rule, preparing for subsequent anomaly identification and deviation handling. The preliminary reliability comprehensive score data is imported into the quality control module, calculating the mean and standard deviation of all container scores. Further, the "standardized deviation" between each score point and the mean is calculated, i.e., the deviation value divided by the standard deviation. When the absolute value of the deviation for a container is greater than twice the standard deviation, the system marks that container as a "statistical outlier."Subsequently, the platform automatically reviews the calculation paths of its stability and fatigue scores. For example, if an abnormally high stability score is found, it may be due to missing historical data for a certain period of use, failing to identify negative behaviors. The system will trigger a recalculation mechanism to remove abnormal samples or correct them with data from other periods, ultimately generating a set of verified comprehensive reliability scores to ensure the scoring system is robust and objectively reflects the container's status. Based on the verified comprehensive scores, a reliability level classification standard is constructed. For example, a score ≥90 is defined as Grade A (high reliability), 70 ≤ score <90 is defined as Grade B (medium reliability), and less than 70 is defined as Grade C (low reliability). The system automatically tags all containers and records them in a "Container Reliability Ranking Table," which includes the container ID, comprehensive score value, level label, and score ranking number. Taking container OX-005148 as an example, its verified score is 95.2, classifying it as Grade A and ranking it 6th. This ranking table is directly used by industrial gas companies for reuse decision-making and scheduling, ensuring that high-reliability containers are prioritized for deployment in hospital clinics, avoiding potential accidents caused by uneven use or excessive fatigue, and improving the overall safety level of the medical oxygen supply chain.

[0084] Preferably, step S3 includes the following steps:

[0085] Step S31: Calculate the statistical characteristic parameters of the reliability distribution based on the container reliability ranking table to obtain the reliability statistical benchmark data;

[0086] The data processing system of this invention first reads the comprehensive score data of all containers from the generated container reliability ranking table, and uses statistical methods to calculate the key statistical characteristic parameters of the score data, such as mean, variance, median, and quantiles (e.g., 25th and 75th percentiles). These statistics constitute the distribution characteristics of the reliability of this batch of containers, used to describe the overall reliability performance of the container group. The system automatically completes this statistical process without manual intervention, and the statistical results serve as the reliability statistical benchmark data, providing data support for subsequent classification. For example, after statistically analyzing the score data of 1000 containers, the average score is 85 points, and the standard deviation is 10 points.

[0087] Step S32: Determine the boundary thresholds for reliability grading based on reliability statistical benchmark data to obtain grading threshold parameters;

[0088] Based on the statistical characteristics obtained in the first step, this embodiment of the invention automatically determines the boundary thresholds for reliability levels by setting reasonable grading principles (e.g., using the mean and standard deviation for stratification, or employing the quartile method). For example, the system can define containers with reliability scores higher than the mean plus one standard deviation as high-level, containers between the mean and one standard deviation as medium-level, and containers lower than the mean minus one standard deviation as low-level. This process is automated, dynamically adjusting parameters to adapt to the score distribution of different batches of containers, generating specific numerical values ​​as grading threshold parameters, such as a high-level threshold of 95 points, a medium-level threshold of 75 to 95 points, and a low-level threshold of below 75 points.

[0089] Step S33: Match the reliability level of each container based on the hierarchical threshold parameter to obtain the container reliability level identifier;

[0090] In this embodiment of the invention, the overall score of each container after verification is compared with the aforementioned grading threshold to determine whether it belongs to a high, medium, or low level. For example, the overall score of container OX-005148 is 92 points, falling within the medium level range, and is therefore automatically marked as a "medium level" container by the system. This matching process is executed in batches, and the system assigns a level identifier to all containers and saves the results to the container information database for scheduling and subsequent management, realizing dynamic updating and tracking of container reliability information.

[0091] Step S34: Establish a reuse category allocation mapping relationship based on the container reliability level identifier to obtain the reuse category mapping rule. Specifically, the reuse category mapping rule is to mark high reliability containers as the direct reuse category, medium reliability containers as the post-detection reuse category, and low reliability containers as the maintenance category or the scrap category.

[0092] Based on industry safety standards and operational experience, this invention maps different reliability levels to specific container reuse categories, forming mapping rules. For example, "high-level" containers, due to their superior performance, can directly enter the reuse process; "medium-level" containers must first undergo IoT sensor detection and manual verification before reuse is permitted; and "low-level" containers must undergo maintenance and upkeep, with severe cases requiring disposal. This mapping relationship is embedded in the management system's rule base and continuously optimized in conjunction with safety compliance requirements to ensure a balance between container safety and economic benefits.

[0093] Step S35: Assign a reuse category to each container based on the reuse category mapping rule to obtain the container reuse category assignment result;

[0094] This invention reads the reliability level identifier of a container and assigns a specific reuse category label to each container according to mapping rules. For example, the medium-level identifier OX-005148 corresponds to "reuse after detection." This category information, along with the container identifier, is entered into the container management system to track the subsequent flow path of the container and issue execution instructions. All containers undergo this automatic allocation step, avoiding inconsistencies caused by subjective human judgment and improving allocation efficiency and accuracy.

[0095] Step S36: Based on the container reuse category allocation results, statistically analyze the distribution of container quantity in each category, and perform supply and demand balance checks and adjustments to obtain container reuse scheduling data.

[0096] This invention, based on the current container reuse category allocation results, counts the number of containers categorized as directly reusable, reusable after testing, maintained, and scrapped, and performs supply and demand balance analysis in conjunction with hospital oxygen demand plans and delivery cycles. If the number of directly reusable containers is insufficient to meet short-term supply demand, the system automatically triggers an adjustment mechanism, prioritizing increasing the frequency of rapid testing for reusable after testing containers, or optimizing the repair process for maintained containers, ensuring supply chain stability. Simultaneously, the system automatically generates the latest container reuse scheduling data as the basis for the next cycle's delivery and inventory management, achieving intelligent allocation and ensuring the efficient and safe distribution of medical oxygen resources.

[0097] Of particular importance, step S36 includes the following steps:

[0098] Step S361: Based on the container reuse category allocation results, count the number of containers in each reuse category and calculate the proportion of each category of containers to the total number of containers to obtain reuse category proportion analysis data;

[0099] The management system in this embodiment of the invention automatically reads the reuse category allocation results of all containers from the container information database and uses a batch data processing algorithm to count the number of containers in four categories: direct reuse, reuse after inspection, maintenance, and scrapping. Subsequently, the system calculates the proportion of each category based on the total number of all containers; for example, direct reuse accounts for 40%, reuse after inspection for 35%, maintenance for 20%, and scrapping for 5%. These proportions constitute reuse category proportion analysis data, used for subsequent quantitative analysis of supply and demand balance. The entire process requires no manual intervention, supports real-time dynamic updates, and ensures the transparency and accuracy of container resource usage.

[0100] Step S362: Based on the reuse category ratio analysis data, conduct a supply and demand balance assessment to obtain supply and demand balance status assessment data. Specifically, the supply and demand balance assessment involves identifying existing category capacity mismatches.

[0101] This invention, based on the current demand plans of industrial gas companies for different types of containers in medical institutions, and combined with real-time inventory and reuse category ratio analysis data, uses a supply-demand matching algorithm to automatically assess whether the supply capacity of each type of container meets the demand. If the system detects that the quantity of a certain type of container (e.g., the directly reused type) is insufficient to meet peak oxygen demand, it will identify a risk of capacity shortage in that category and generate a mismatch warning. The supply-demand balance assessment data records in detail the percentage difference between supply and demand for each category and the potential risk level, providing a scientific basis for scheduling optimization.

[0102] Step S363: Based on the supply and demand balance assessment data, the container categories with mismatches are reallocated to obtain the adjusted reuse category allocation data;

[0103] In this embodiment of the invention, for categories with supply-demand imbalances, the system automatically invokes the reuse category adjustment module. Based on rules, it prioritizes increasing the utilization rate of containers in the medium reliability category (reusable after inspection), such as increasing inspection frequency or extending service life. It also dynamically adjusts some maintenance category containers to be temporarily converted to the reusable after inspection category, alleviating the supply-demand imbalance. This adjustment is based on real-time IoT monitoring data, combined with container reliability and fatigue status, ensuring maximum utilization efficiency under safe conditions. The adjustment results are written to the database in real time, forming the latest reuse category allocation data, enabling intelligent scheduling that quickly responds to changes in medical oxygen demand.

[0104] Step S364: Based on the reuse category allocation data, generate a processing priority sort for each container to obtain container reuse scheduling data including container identifier, reuse category and processing priority.

[0105] This invention, in its embodiments, considers the container's reuse category, reliability score, fatigue accumulation value, and the urgent needs of the current medical institution, employing a multi-factor weighted ranking algorithm to generate a specific processing priority for each container. Containers with higher priority will be prioritized for filling and distribution, while those with lower priority will be scheduled for maintenance or standby. The scheduling data includes the container's unique identifier, its corresponding reuse category, and the calculated processing priority value, achieving closed-loop management from container data collection to actual distribution operations. This ensures a safe and efficient supply of medical oxygen, with the entire process automated to minimize human error.

[0106] Preferably, step S4 includes the following steps:

[0107] Step S41: Based on container reuse scheduling data, establish a multi-queue parallel processing framework, create four independent processing queues: direct reuse, detection reuse, maintenance, and scrapping, and obtain the parallel processing queue framework;

[0108] This invention utilizes scheduling data from an IoT management platform as a foundation. The system automatically analyzes container reuse category information and establishes four independent and parallel processing queues, corresponding to direct reuse (rapid filling and delivery without additional detection), reuse after detection (requiring status verification), maintenance (requiring equipment repair and performance recovery), and scrapping (requiring safe disposal and resource recycling). This multi-queue framework, through advanced queue management algorithm design, supports asynchronous parallel processing, effectively improving container processing efficiency and ensuring that different categories of containers are rationally scheduled according to their characteristics and process requirements. Each queue in the queue framework can be configured with priority parameters and resource allocation ratios to ensure timely processing of high-priority containers. The system dynamically adjusts the processing capacity of each queue based on the number of containers and real-time demand.

[0109] Step S42: Allocate each container to the parallel processing queue framework according to the container reuse scheduling data to obtain the initial queue allocation state;

[0110] This invention automatically reads the reuse category and processing priority of each container, imports container identifiers and their scheduling data in batches from the database, and accurately maps them to the corresponding processing queues. During the allocation process, the platform inserts the container number into the data structure of the corresponding queue based on the container's category label and marks the processing status as "pending processing". To avoid congestion in a single queue, the system adopts a dynamic load balancing algorithm, monitors the number of containers and processing rate in each queue, and automatically adjusts the allocation of newly enqueued containers to form the initial queue allocation state. This process is completed fully automatically, ensuring accurate and efficient container allocation and laying the foundation for subsequent classification and processing.

[0111] Step S43: Based on the initial queue allocation state, perform classification processing operations on containers of each category respectively, quickly fill containers of the direct reuse category, verify the status of containers of the reuse category after detection, repair and restore the equipment of containers of the maintenance category, and recycle the resources of containers of the scrap category, to obtain direct reuse processing result data, dynamic reclassification result data, maintenance processing result data and scrap processing result data;

[0112] In this embodiment of the invention, containers in each processing queue enter their respective process flows. Reusable containers are quickly located using IoT automatic identification technology, enabling efficient filling at automated filling stations. The system monitors filling pressure and time parameters in real time to ensure a safe and accurate filling process. Reusable containers, after inspection, enter an automated inspection line. Integrated pressure sensors, leak detection devices, and intelligent diagnostic algorithms comprehensively verify the container status. Inspection results are automatically fed back; abnormal containers are dynamically reclassified to maintenance or scrap categories, generating dynamic reclassification data. Maintenance containers undergo structural repair, seal replacement, and pressure performance verification via equipment maintenance robots, updating their status after performance is restored. Scrap containers enter the safe dismantling and resource recycling stage, with all operational data uploaded to the system in real time. After each category is processed, corresponding result data is generated for overall process analysis and optimization.

[0113] Step S44: Summarize the direct reuse processing result data, dynamic reclassification result data, maintenance processing result data, and scrapping processing result data into a processing execution report, thereby realizing the reuse allocation of recycled medical oxygen storage containers.

[0114] This invention collects and summarizes all operational data from four processing queues, combining container identifiers, processing times, status change records, quality inspection results, and maintenance logs to generate a standardized processing execution report. The report details the final processing status, flow path, and time points for each container, providing management with a fully transparent overview of container allocation and utilization. Furthermore, the report supports export and API calls, facilitating integration with supply chain management systems and customer service platforms to achieve a closed-loop management system for the reuse and allocation of recycled containers. This process requires no manual intervention, ensuring data consistency and real-time updates, thereby improving the operational efficiency and safety of industrial gas companies.

[0115] Of particular importance, step S43 includes the following steps:

[0116] Step S431: Perform a fast filling operation on the containers of the directly reused queue in the initial queue allocation state, and record the operation results and exceptions to obtain the direct reuse processing result data;

[0117] This invention utilizes an IoT system to automatically identify container identifiers in the direct reuse queue and employs intelligent filling equipment to efficiently fill these containers. During the filling process, the equipment collects pressure sensor data, filling time, and flow information in real time, automatically determining whether the filling has reached the preset pressure standard and safety parameters. If any abnormality is detected, such as abnormally prolonged filling time, pressure failure, or leakage alarm, the system immediately generates an abnormality alarm and records detailed abnormality information. All operation data and abnormality records are automatically stored in a database, forming direct reuse processing result data, ensuring the traceability of the filling status of each container and supporting subsequent analysis.

[0118] Step S432: Perform status verification on the containers in the reuse queue detected in the initial queue allocation state. Transfer the verified containers to the filling process, and transfer the unverified containers to the maintenance queue or scrap queue to obtain dynamic reclassification result data.

[0119] This invention addresses containers in the reuse queue. The system automatically invokes an integrated intelligent diagnostic module, which utilizes historical sensor data and real-time on-site detection results, including leak detection, pressure holding capacity testing, and structural integrity scanning, to comprehensively review the container's status. The review process employs a machine learning model combined with a rule engine to automatically determine whether the container meets the continued-use criteria. Containers meeting the criteria are automatically transferred to the intelligent filling line, while those not meeting the criteria are automatically categorized into a maintenance queue (e.g., small-scale leaks or minor structural damage) or a scrap queue (e.g., severe structural damage or safety hazards) based on the type and severity of the fault. The entire review and classification process is automated, with dynamic reclassification results updated to the management platform in real time.

[0120] Step S433: Update the container list of the maintenance queue and the scrap queue in real time based on the dynamic reclassification result data to obtain the updated processing queue status;

[0121] This invention, through real-time monitoring of dynamic reclassification results, automatically updates the information of newly added containers to the maintenance and scrapping queues into the corresponding queue databases. The updates include key fields such as container identifier, classification timestamp, verification test results, and anomaly descriptions. The system periodically executes inventory synchronization tasks to ensure that the maintenance and scrapping queues reflect the latest container distribution, and uses historical data analysis for capacity estimation and resource allocation. The updated processing queue status interface allows administrators to remotely view and adjust the status, ensuring the orderly conduct of subsequent maintenance and scrapping operations.

[0122] Step S434: Based on the updated processing queue status, perform device repair and performance recovery processing on the containers in the maintenance queue to obtain maintenance processing result data;

[0123] In this embodiment of the invention, containers in the maintenance queue enter an automated repair workshop established by an industrial gas company. Using robotic repair arms and specialized repair equipment, critical components are replaced, seals are updated, and surfaces are cleaned. Simultaneously, pressure testing instruments and leak detection equipment verify the repair effectiveness. All repair steps and test data are collected in real time by IoT sensors, automatically generating detailed maintenance results data, including repair items, test results, and performance recovery indicators. After repair, the container status is updated to "usable" or "requires re-inspection," and relevant data is automatically uploaded to the management platform, ensuring transparency and data integrity in the maintenance process.

[0124] Step S435: Perform resource reclamation on the containers in the scrap queue based on the updated processing queue status to obtain scrap processing result data.

[0125] In this embodiment of the invention, containers in the disposal queue enter a dedicated resource recycling process according to the environmental and safety standards set by the industrial gas company. This process uses automated dismantling equipment to structurally break down the containers, recovering steel and sealing materials. Simultaneously, sensors monitor the emission of harmful gases and the safe handling of residual oxygen during the dismantling process. The system records dismantling time, recovery rate, and safety monitoring data throughout the process, generating disposal result data for subsequent environmental compliance reports and material reuse statistics. This process ensures maximum resource utilization and complies with industry safety requirements.

[0126] Figure 3 The core application of this application is a complete portable medical oxygen supply device. This device constitutes the basic monitoring and management unit in the technical solution of this invention. By intelligently modifying this device, precise monitoring and intelligent management of the entire medical oxygen supply network can be achieved. This includes:

[0127] The storage container body 101 is a standard high-pressure gas cylinder, painted blue in accordance with international labeling standards for medical oxygen. In this invention, the storage container body 101, after intelligent modification, becomes a data collection unit. The storage container body 101 integrates various sensor devices on its surface, including a weight sensor for monitoring gas consumption, a temperature sensor for recording environmental impacts, a vibration sensor for detecting physical shocks during transportation and use, and an RFID tag for identification and data storage. The data collected by these sensors constitutes the basic information source for the behavior analysis module to establish usage characteristics.

[0128] Pressure regulator 102: This is the core control component of the entire gas supply system, responsible for reducing the pressure of the high-pressure gas inside the container to a safe pressure range suitable for medical use. In the intelligent solution of this invention, the traditional mechanical pressure regulator is upgraded to an intelligent regulator integrating digital monitoring functions. The pressure regulator 102 integrates a digital pressure sensor to accurately measure changes in input and output pressure; a flow sensor records the rate and total volume of gas flow; and a timestamp recorder accurately marks the start and end times of each use. This data provides crucial information for establishing container usage behavior characteristics and calculating cumulative fatigue values.

[0129] Safety monitoring device 103: An advanced safety monitoring module is installed in the intelligent system of this invention. The safety monitoring device 103 continuously monitors key parameters such as system pressure, flow rate, and temperature. When an abnormality is detected, it immediately issues an alarm and records detailed fault information. This device has predictive fault diagnosis capabilities, predicting potential problems by analyzing the changing trends of operating parameters, and providing important status data for the reliability assessment module.

[0130] Flow meter 104: Accurately measures the flow rate of the gas passing through and displays the instantaneous flow rate value on the pipe wall scale. In the intelligent transformation of this invention, the traditional mechanical reading flow meter is upgraded to a digital intelligent flow meter. Flow meter 104 accurately measures the flow rate and records detailed usage pattern data, identifies different usage scenarios, records the flow curve changes for each use, and analyzes the temporal distribution characteristics of usage frequency. This flow data constitutes the core information source for establishing usage frequency patterns and analyzing usage behavior characteristics.

[0131] Humidification buffer container 105: Real-time monitoring of water level changes and recording the correlation between oxygen throughput and water consumption. A temperature sensor monitors internal temperature changes and analyzes the impact of gas flow on humidification. A flow sensor measures the gas flow rate through the container, providing supplementary data for calculating total oxygen usage.

[0132] When the medical oxygen device is activated to begin treatment, the entire intelligent monitoring network immediately comes into operation. First, the weight sensor integrated on the storage container body 101 detects minute changes in the container's weight, directly reflecting oxygen consumption. Simultaneously, the temperature sensor on the storage container body 101 records the impact of ambient temperature on the container, while the vibration sensor monitors for potential physical disturbances during use. This foundational data provides crucial environmental context for subsequent usage behavior analysis. At the gas flow control level, the pressure regulator 102 plays a central coordinating role. As oxygen flows from the storage container body 101 to the patient, the digital pressure sensor inside the pressure regulator 102 immediately begins monitoring the dynamic changes in input and output pressure. This detailed recording of pressure changes forms the crucial pressure drop curve data in our technical solution. Simultaneously, the flow sensor on the pressure regulator 102 accurately measures the instantaneous rate of gas flow, while the timestamp recorder ensures accurate time stamping for each data point.

[0133] The flow meter 104 plays a crucial role in accurate metering throughout the entire collaborative system. As oxygen passes through the flow meter 104, its built-in intelligent sensors not only measure the instantaneous flow rate but also analyze the dynamic patterns of flow rate changes. The flow meter 104 can identify different modes of continuous and intermittent use, record the flow rate curve characteristics for each use, and analyze the distribution of usage frequency over time. This detailed flow data is cross-validated with the pressure data provided by the pressure regulator 102, ensuring the accuracy of the usage behavior analysis.

[0134] The humidification buffer container 105 provides important supplementary monitoring functions in the collaborative mechanism. When oxygen is humidified through the humidification buffer container 105, a level sensor inside the container monitors water level changes, establishing a quantitative relationship between oxygen throughput and water consumption. A temperature sensor in the humidification buffer container 105 analyzes the impact of gas flow on the humidification effect, while its flow sensor provides cross-validation data for calculating the total oxygen usage.

[0135] The safety monitoring device 103 plays a crucial role in system supervision throughout the entire collaborative operation process. The safety monitoring device 103 continuously receives various operating parameters from the storage container body 101, pressure regulator 102, flow meter 104, and humidification buffer container 105, and performs real-time comprehensive analysis. When the safety monitoring device 103 detects any parameter exceeding the normal range, it immediately records detailed information about the anomaly and issues a corresponding alarm. More importantly, by analyzing the changing trends of data from each component, the safety monitoring device 103 can predict potential system problems, providing forward-looking condition assessment data for the reliability assessment module.

[0136] At the data transmission and processing level, an efficient information flow network is formed among the various components. The RFID tag on the main body of the storage container 101 is responsible for device identification and basic data storage, ensuring that the data of each device can be accurately attributed. Real-time data collected by all components is uniformly transmitted to the cloud data processing center through the built-in wireless communication module. When the network connection is not working, the local storage function of each component ensures that the data is not lost, and the data is automatically uploaded after the network is restored.

[0137] When a use is completed, the entire collaborative system enters the data integration phase. All data collected by the storage container body 101, pressure regulator 102, flow meter 104, humidification buffer container 105, and safety monitoring device 103 are integrated and processed to form a complete data record for that use. This data includes usage duration, gas consumption, pressure change curves, flow change patterns, environmental impact factors, and safety status records.

[0138] The present invention also provides an Internet of Things (IoT)-based medical oxygen storage and distribution system for executing the above-described IoT-based medical oxygen storage and distribution method, wherein the IoT-based medical oxygen storage and distribution system comprises:

[0139] The behavior analysis module is used to acquire usage data of the recovered medical oxygen storage containers; it uses the usage data to analyze the pressure drop curve and usage frequency pattern of each container when oxygen is drawn, and establishes usage behavior characteristics; by comparing the stability differences of the usage behavior characteristics of different containers, it obtains container stability classification data.

[0140] The reliability assessment module is used to statistically analyze the cumulative number of uses and load intensity changes of each container based on the container stability classification data; calculate the cumulative fatigue value of each container based on the cumulative number of uses and load intensity changes; combine the cumulative fatigue value with the container stability classification data to assess the container reliability and generate a container reliability ranking table.

[0141] The intelligent scheduling module is used to perform reliability matching of medical oxygen storage containers according to the reliability ranking of containers, and to divide them into reuse categories to obtain container reuse scheduling data. Specifically, the reuse categories are marked as high reliability containers as direct reuse category, medium reliability containers as reuse category after inspection category, and low reliability containers as maintenance category or scrap category.

[0142] The classification execution module is used to perform classification processing and reallocation based on container reuse scheduling data. It performs rapid filling of containers in the direct reuse category, status verification of containers in the reuse category after testing, equipment repair and performance restoration of containers in the maintenance category, and resource recycling of containers in the scrap category, thereby realizing the reuse allocation of recycled medical oxygen storage containers.

[0143] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0144] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A medical oxygen storage and distribution method based on the Internet of Things, characterized by, Includes the following steps: Step S1: Obtain usage data of the recycled medical oxygen storage containers; By analyzing the pressure drop curves and usage frequency patterns of each container when oxygen is drawn, usage behavior characteristics can be established. By comparing the stability differences in the usage behavior characteristics of different containers, container stability grading data is obtained. Step S2: Based on the container stability classification data, calculate the cumulative number of uses and load intensity changes of each container; calculate the cumulative fatigue value of each container based on the cumulative number of uses and load intensity changes; combine the cumulative fatigue value with the container stability classification data to evaluate the container reliability and generate a container reliability ranking table. Step S2, which calculates the cumulative fatigue value of each vessel based on the cumulative number of uses and changes in load intensity, includes: The utilization rate percentage is calculated by dividing the cumulative number of times each container is used by its designed service life, thus obtaining the container utilization rate data. Risk classification is performed based on container usage data. When the usage data exceeds 80%, it is marked as high usage and assigned a risk coefficient of 1.

5. When the usage data is below 80%, it is marked as standard usage and assigned a risk coefficient of 1.0, thus obtaining the usage risk coefficient. The load intensity ratio data is obtained by calculating the ratio of the average working pressure to the design standard pressure based on the historical load intensity data of the container. Overload risk assessment is performed based on the load intensity ratio. When the load intensity ratio exceeds 1.2, it is marked as overloaded use and a penalty coefficient of 2.0 is assigned. When the load intensity ratio is below 1.2, it is marked as normal use and a penalty coefficient of 1.0 is assigned. The overload coefficient is then obtained. Based on the analysis of load intensity changes, load change trends and fluctuation characteristics are determined. When the load shows an upward trend, a trend coefficient of 1.3 is assigned, and when the load fluctuation exceeds 50% of the average value, a fluctuation coefficient of 1.4 is assigned, thus obtaining the load change adjustment coefficient. The comprehensive fatigue calculation is based on the utilization rate risk coefficient, overload coefficient and load change adjustment coefficient. The container utilization rate data is multiplied by the utilization rate risk coefficient and then by 60% weight, plus the load intensity ratio multiplied by the overload coefficient and then by 25% weight, plus the load change adjustment coefficient multiplied by 15% weight, to obtain the comprehensive fatigue cumulative index. Risk classification is performed based on the comprehensive fatigue accumulation index to obtain the container fatigue accumulation value; Step S2, which combines the fatigue accumulation value with the vessel stability grading data to assess vessel reliability, includes: Based on the stability index values ​​in the container stability grading data, the stability index values ​​are multiplied by a standardization coefficient and then subjected to interval mapping to transform the value range of the stability index to a scoring range of 0-100, thus obtaining the container stability score. The container fatigue resistance score is obtained by performing reverse numerical transformation based on the cumulative fatigue value of the container and multiplying the reciprocal of the cumulative fatigue value by a preset standardization factor. The relative weights are determined by calculating the ratio of the standard deviations of the two types of scores, namely, the container stability score and the fatigue resistance score, and the dynamic weight allocation parameters are obtained. The container stability score and fatigue resistance score are weighted and averaged according to the dynamic weight allocation parameters to obtain a preliminary comprehensive reliability score. Based on the preliminary comprehensive reliability score, the standardized deviation of each container score from the group average is calculated. When the absolute value of the standardized deviation exceeds twice the standard deviation, it is marked as a statistical outlier. The outliers are then verified by data tracing and recalculated to obtain the verified comprehensive reliability score. The reliability levels are divided based on the comprehensive reliability score after verification, resulting in a container reliability ranking table. Step S3: Perform reliability matching on medical oxygen storage containers according to container reliability ranking, and divide them into reuse categories to obtain container reuse scheduling data. Specifically, the reuse categories are divided into high reliability containers as direct reuse category, medium reliability containers as reuse category after inspection, and low reliability containers as maintenance category or scrap category. Step S4: Based on the container reuse scheduling data, perform classification processing and reallocation, quickly fill containers in the direct reuse category, verify the status of containers in the reuse category after testing, repair and restore the performance of containers in the maintenance category, and recycle the resources of containers in the scrap category, thereby realizing the reuse allocation of recycled medical oxygen storage containers.

2. The IoT based medical oxygen storage dispensing method as claimed in claim 1 wherein, Step S1 includes the following steps: Step S11: Synchronously read the identification code of the recycled medical oxygen storage container and the stored usage history data through the Internet of Things sensor network, wherein the usage history data includes the time of release, the location of use, the time of recycling, and the stored sensor data; Step S12: Arrange the usage history data and identification code in chronological order to form a continuous usage process record, and obtain container identity association data; Step S13: Perform integrity verification and data cleaning on the container identity association data, check for missing data, incorrect timestamps or sensor malfunction records, and mark the time periods with incomplete data to obtain usage history data; Step S14: Analyze the pressure drop curves and usage frequency patterns of each container when oxygen is drawn using the usage process data to establish usage behavior characteristics; Step S15: By comparing the stability differences in the usage behavior characteristics of different containers, container stability grading data is obtained.

3. The IoT based medical oxygen storage dispensing method as claimed in claim 2, wherein, Step S14 includes the following steps: Step S141: Extract the pressure change time series from the historical data, identify the pressure drop events corresponding to the customer's oxygen intake operation, and record the starting point, rate of decrease and ending point of each pressure drop by continuously monitoring the downward trend and magnitude of the pressure value to obtain the pressure drop curve. Step S142: Based on the usage process data, statistically analyze the usage frequency and time distribution characteristics of each container, calculate the average number of uses per unit time, the regularity of the usage time interval, and the periodic changes in usage intensity to obtain the usage process time characteristics; Step S143: Generate the usage frequency pattern of each container by analyzing the concentration and dispersion of the usage process time characteristics; Step S144: Extract features and quantify the pressure drop curve and usage frequency pattern. Extract the stability of the drop rate, the consistency of the response time, and the pressure holding capability from the pressure drop curve. Extract the frequency regularity, the adaptability of usage intensity, and the uniformity of time distribution from the usage frequency pattern, thereby establishing usage behavior characteristics that describe the performance of the container.

4. The medical oxygen storage and distribution method based on the Internet of Things according to claim 3, characterized in that, Step S15 includes the following steps: Step S151: Perform a horizontal comparison of the usage behavior characteristics of all recycling containers, calculate the numerical differences and distribution range of the usage behavior characteristic parameters of each container, and obtain the outlier distribution matrix data of container usage behavior. Step S152: Identify stable and unstable container groups based on the outlier distribution matrix data, quantify the deviation of each container from the benchmark of the stable container group, and obtain the stability difference; Step S153: Establish a container stability grading standard based on the numerical distribution of stability differences, and assign a corresponding identification code to each grade to obtain container stability grading data.

5. The IoT based medical oxygen storage dispensing method as claimed in claim 4, wherein, Step S2 involves statistically analyzing the cumulative usage counts and load intensity changes of each container based on the container stability classification data, including: Based on the container identification information in the container stability classification data, extract the outbound and return records of each container since it was put into use, and classify and organize them according to the container number to form an individual usage file for each container. Based on the total number of times each container is dispatched according to individual usage records, a container dispatch statistics table is obtained. Based on the container outbound frequency statistics table, calculate the distribution of usage frequency of each container in different time periods, and perform cumulative summation. Add the usage frequency of each time period step by step in chronological order to calculate the cumulative usage frequency from the time of use to the current time, and obtain the cumulative usage frequency of each container. Extract pressure load information for each use process from individual usage records, including the filling pressure value at each time of departure, the working pressure range during use, and the remaining pressure value at the time of recovery. Calculate the average working pressure, maximum working pressure, and pressure change range of each container in each use to obtain historical load intensity data of the container. By comparing the differences in average working pressure over different periods based on historical load intensity data of the container, the trend of load intensity can be identified, and the change in load intensity can be obtained.

6. The IoT based medical oxygen storage dispensing method as claimed in claim 5, wherein, Step S3 includes the following steps: Step S31: Calculate the statistical characteristic parameters of the reliability distribution based on the container reliability ranking table to obtain the reliability statistical benchmark data; Step S32: Determine the boundary thresholds for reliability grading based on reliability statistical benchmark data to obtain grading threshold parameters; Step S33: Match the reliability level of each container based on the hierarchical threshold parameter to obtain the container reliability level identifier; Step S34: Establish a reuse category allocation mapping relationship based on the container reliability level identifier to obtain the reuse category mapping rule. Specifically, the reuse category mapping rule is to mark high reliability containers as the direct reuse category, medium reliability containers as the post-detection reuse category, and low reliability containers as the maintenance category or the scrap category. Step S35: Assign a reuse category to each container based on the reuse category mapping rule to obtain the container reuse category assignment result; Step S36: Based on the container reuse category allocation results, statistically analyze the distribution of container quantity in each category, and perform supply and demand balance checks and adjustments to obtain container reuse scheduling data.

7. The IoT based medical oxygen storage dispensing method as claimed in claim 6, wherein, Step S4 includes the following steps: Step S41: Based on container reuse scheduling data, establish a multi-queue parallel processing framework, create four independent processing queues: direct reuse, detection reuse, maintenance, and scrapping, and obtain the parallel processing queue framework; Step S42: Allocate each container to the parallel processing queue framework according to the container reuse scheduling data to obtain the initial queue allocation state; Step S43: Based on the initial queue allocation state, perform classification processing operations on containers of each category respectively, quickly fill containers of the direct reuse category, verify the status of containers of the reuse category after detection, repair and restore the equipment of containers of the maintenance category, and recycle the resources of containers of the scrap category, to obtain direct reuse processing result data, dynamic reclassification result data, maintenance processing result data and scrap processing result data; Step S44: Summarize the direct reuse processing result data, dynamic reclassification result data, maintenance processing result data, and scrapping processing result data into a processing execution report, thereby realizing the reuse allocation of recycled medical oxygen storage containers.

8. An Internet of Things based medical oxygen storage and distribution system characterized in that, For executing the IoT-based medical oxygen storage and distribution method as described in claim 1, the IoT-based medical oxygen storage and distribution system comprises: The behavior analysis module is used to acquire usage data of the recovered medical oxygen storage containers; it uses the usage data to analyze the pressure drop curve and usage frequency pattern of each container when oxygen is drawn, and establishes usage behavior characteristics; by comparing the stability differences of the usage behavior characteristics of different containers, it obtains container stability classification data. The reliability assessment module is used to statistically analyze the cumulative number of uses and load intensity changes of each container based on the container stability classification data; calculate the cumulative fatigue value of each container based on the cumulative number of uses and load intensity changes; combine the cumulative fatigue value with the container stability classification data to assess the container reliability and generate a container reliability ranking table. The intelligent scheduling module is used to perform reliability matching of medical oxygen storage containers according to the reliability ranking of containers, and to divide them into reuse categories to obtain container reuse scheduling data. Specifically, the reuse categories are marked as high reliability containers as direct reuse category, medium reliability containers as reuse category after inspection category, and low reliability containers as maintenance category or scrap category. The classification execution module is used to perform classification processing and reallocation based on container reuse scheduling data. It performs rapid filling of containers in the direct reuse category, status verification of containers in the reuse category after testing, equipment repair and performance restoration of containers in the maintenance category, and resource recycling of containers in the scrap category, thereby realizing the reuse allocation of recycled medical oxygen storage containers.

Images