A medical liquid oxygen storage tank remote monitoring and gas supply early warning method and system based on the Internet of Things and a medium

Abstract

This application relates to a method, system, and medium for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things (IoT), belonging to the field of monitoring and early warning technology. The method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks includes: monitoring the operating status of the liquid oxygen storage tank, collecting multi-dimensional liquid oxygen storage data, constructing and transmitting liquid oxygen data packets; parsing historical liquid oxygen parameter packets, combining them with tank geometric parameters, calculating the remaining liquid oxygen mass, forming a liquid oxygen time-series data curve, and generating a multi-level threshold early warning model; detecting liquid oxygen data packets according to the multi-level threshold early warning model, calculating the available liquid oxygen time, and generating a liquid oxygen replenishment early warning work order; scheduling liquid oxygen transport tanks to replenish liquid oxygen storage tanks to the target liquid oxygen limit according to the liquid oxygen replenishment early warning work order; through the multi-level early warning mechanism, safety hazards such as oxygen deficiency and supply interruption and pressure exceeding the standard are effectively prevented, ensuring the safety and continuity of clinical oxygen use, and improving the intelligent level of hospital logistics management and emergency response speed.

Description

Technical Field

[0001] This application relates to the field of monitoring and early warning technology, and in particular to a method, system and medium for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things. Background Technology

[0002] In modern hospitals, liquid oxygen is an indispensable life support resource for surgical anesthesia, intensive care, and emergency treatment. Traditional liquid oxygen supply management mainly relies on manual meter reading and periodic inspections, where equipment department personnel periodically visit the storage tank area to read pressure gauge and level gauge data. This model has significant time lags and safety hazards.

[0003] Existing patents disclose an intelligent monitoring and alarm system for a hospital liquid oxygen station, comprising a liquid oxygen station chamber, a monitoring terminal, and a mobile APP. The inner bottom wall of the liquid oxygen station chamber is fixedly connected to two supports, the inner walls of which together hold a liquid oxygen storage tank. A pressure gauge is fixedly embedded in the left end of the liquid oxygen storage tank, and a pressure monitoring sensor and a liquid oxygen level sensor are fixedly embedded in the upper surface of the tank. This invention, by incorporating a pressure monitoring sensor, a liquid oxygen level sensor, a liquid oxygen monitoring sensor, a signal acquisition unit, a monitoring main control chassis, a signal transmitter, an audible and visual alarm, an infrared camera, a monitoring terminal, and a mobile APP, can monitor the pressure, level, and other parameters of the liquid oxygen storage tank in case of oxygen leakage and provide timely audible and visual alarms. It also enables remote real-time monitoring and alarming of oxygen leakage in the liquid oxygen storage tank, facilitating timely maintenance and handling by staff and addressing the problem of delayed detection of oxygen leaks leading to dangerous accidents such as fires.

[0004] The existing technical solutions mentioned above have the following drawbacks: 1. The existing system cannot monitor the consumption rate of liquid oxygen in real time. When encountering public health emergencies or peak surgical periods, there is a risk of insufficient or even interrupted oxygen supply, which may endanger the lives of patients; 2. Traditional monitoring devices often operate independently, and the data cannot be exchanged with the hospital's logistics management system, resulting in a lack of data support for scheduling decisions and low efficiency in emergency response. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this application is to provide a method, system, and medium for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things. Through a multi-level early warning mechanism, it can effectively prevent safety hazards such as oxygen deficiency and supply interruption and pressure exceeding the standard, ensure the safety and continuity of clinical oxygen use, improve the intelligence level of hospital logistics management and emergency response speed, and eliminate the blind spots of manual inspection.

[0006] This was achieved using the following technical solutions: In a first aspect, this application provides a method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things, including: Monitor the operating status of liquid oxygen storage tanks, collect multi-dimensional liquid oxygen storage data, and construct and transmit liquid oxygen data packages; Analyze historical liquid oxygen parameter packages, combine them with tank geometric parameters, calculate the remaining liquid oxygen mass, generate liquid oxygen time series data curves, and produce a multi-level threshold early warning model; The system detects liquid oxygen data packets based on a multi-level threshold early warning model, calculates the available time of liquid oxygen, and generates a liquid oxygen replenishment early warning work order. Based on the liquid oxygen replenishment early warning work order, the liquid oxygen transport tank is dispatched to replenish the liquid oxygen storage tank to the target liquid oxygen limit.

[0007] By adopting the above technical solution, by integrating multi-sensor data and tank geometric parameters, establishing a time-series curve, and using a multi-level threshold early warning algorithm to dynamically calculate the remaining available time of liquid oxygen, the system automatically generates replenishment work orders and schedules transport tanks for precise replenishment. This achieves intelligent monitoring of liquid oxygen storage tanks, accurate calculation of remaining mass, and closed-loop automatic replenishment, significantly improving operational safety, early warning timeliness, and tank utilization.

[0008] This application further specifies: monitoring the operating status of the liquid oxygen storage tank, collecting multi-dimensional liquid oxygen storage data, constructing and transmitting liquid oxygen data packets, including: The operating status of the liquid oxygen storage tank is detected according to the preset collection cycle to obtain the initial liquid oxygen pressure value, initial storage liquid level value and initial liquid oxygen flow rate value. The initial liquid oxygen pressure value, initial storage liquid level value, and initial liquid oxygen flow rate value are filtered, denoised, and anomaly removed to obtain the effective liquid oxygen pressure value, effective liquid oxygen level value, and effective liquid oxygen flow rate value. Based on the data collection timestamp, the effective liquid oxygen pressure value, effective liquid oxygen level value, and effective liquid oxygen flow rate value are correlated and converted, and combined with the liquid oxygen storage tank number, a liquid oxygen data package is constructed. The liquid oxygen data packets are compressed and encrypted based on the network bandwidth to obtain encrypted liquid oxygen packets, which are then transmitted and stored in a time-series database.

[0009] By adopting the above technical solutions, multi-dimensional data such as pressure, liquid level and flow rate are processed based on filtering and noise reduction and anomaly removal algorithms. Secure data packets are constructed and stored in the time series database through time series correlation and compression encryption technologies, which significantly improves the accuracy, integrity and transmission and storage efficiency of liquid oxygen monitoring data.

[0010] This application further specifies: parsing historical liquid oxygen parameter packages, combining them with tank geometric parameters, calculating the remaining liquid oxygen mass, generating liquid oxygen time-series data curves, and producing a multi-level threshold early warning model, including: The historical liquid oxygen parameter package was analyzed and its integrity was checked, and the tank number, liquid oxygen level, tank pressure value and instantaneous flow rate value were screened. The time-series database is matched based on the tank number to extract the corresponding tank geometric parameters; The corresponding liquid oxygen volume conversion function is called according to the shape of the storage tank, and the initial liquid oxygen mass is obtained by combining the tank geometric parameters and the liquid oxygen level height. Based on the tank temperature and the standard density table of liquid oxygen, the initial liquid oxygen mass is corrected for density, and the remaining liquid oxygen mass is calculated. The remaining liquid oxygen mass is interpolated temporally based on the collection timestamp to construct a liquid oxygen time-series data curve; Based on the tank pressure and instantaneous flow rate, the liquid oxygen time series data curve is auxiliaryly derived to form a periodic consumption time series curve; The cycle consumption time series curve and liquid oxygen time series data curve are mapped, divided and feature-iterated to determine multi-level liquid oxygen remaining mass thresholds and generate a multi-level threshold early warning model. Multi-stage liquid oxygen residual mass thresholds include: If the remaining mass of liquid oxygen is within the first consumption range, a mandatory replenishment notification will be generated and marked as a first-level mandatory early warning threshold. If the remaining mass of liquid oxygen is within the second consumption range, an emergency replenishment notification will be generated and marked as a level 2 emergency warning threshold. If the remaining mass of liquid oxygen is within the third consumption range, a planned replenishment notification will be generated and marked as a level three safety warning threshold. The warning thresholds for each level are dynamically adjusted based on the periodic consumption rate to form a dynamic hierarchical warning function.

[0011] By adopting the above technical solution, the remaining mass is calculated by analyzing historical liquid oxygen parameters and matching tank geometric parameters, calling the volume conversion function and combining it with density correction, constructing time series curves and integrating pressure and flow rate to assist in the evolution of cycle consumption curves, using feature iteration to determine three-level early warning thresholds, and using the cycle consumption rate to dynamically correct the hierarchical early warning function, the accuracy of liquid oxygen remaining mass prediction, the timeliness of early warning, and the adaptability to different operating conditions are significantly improved.

[0012] This application further specifies a multi-level threshold early warning model, including: Based on the first feature extraction branch, feature decomposition is performed on the periodic consumption time series curve to extract daily average consumption feature values ​​and monthly average consumption feature values; Based on the second feature extraction branch, feature analysis is performed on the liquid oxygen time series data curve to obtain the mass time series feature value and the tank application frequency feature value; Based on the first iteration training branch combined with the local attention mechanism, the daily average consumption feature value, monthly average consumption feature value and mass time series feature value are trained several times to generate the liquid oxygen mass change feature matrix. Based on the second iteration training branch combined with the spatiotemporal attention mechanism, the quality time series feature value, the tank application frequency feature value and the environmental change feature value are trained several times to generate the liquid oxygen storage and call feature matrix. Based on the integrated training branch, the liquid oxygen mass change feature matrix and the liquid oxygen storage and retrieval feature matrix are fused and spliced ​​to generate the liquid oxygen remaining mass weight matrix. Based on the auxiliary mapping branch and the multi-level liquid oxygen remaining mass threshold, the liquid oxygen remaining mass weight matrix is ​​mapped and differentiated to obtain the hierarchical warning weight matrix. Based on the identification optimization branch, the hierarchical warning weight matrix is ​​called to identify the historical liquid oxygen parameters hierarchically, and the hierarchical accuracy is calculated by combining the warning level label. If the hierarchical accuracy is less than the preset accuracy threshold, the hierarchical warning weight matrix is ​​corrected based on the dynamic hierarchical warning function and environmental change factors, and then identified again until the target is met. If not, then time-series fitting is performed on the dynamic hierarchical early warning function and the hierarchical warning weight matrix to generate a multi-level threshold early warning model.

[0013] By adopting the above technical solution, the consumption and storage features are extracted based on the dual-branch attention network, and the remaining mass weight matrix is ​​generated by fusion. After hierarchical mapping and dynamic hierarchical early warning function iterative correction, the accurate construction of the multi-level threshold early warning model for liquid oxygen storage tanks is realized, which significantly improves the reliability and adaptability of remaining mass prediction.

[0014] This application is further configured to: detect liquid oxygen data packets based on a multi-level threshold early warning model, calculate the available liquid oxygen time, and generate a liquid oxygen replenishment early warning work order, including: The liquid oxygen data packet is parsed to obtain the effective liquid oxygen level value and the effective liquid oxygen flow rate value; The effective liquid oxygen level is converted based on the tank's geometric parameters to obtain the effective liquid oxygen volume, and the remaining liquid oxygen mass is calculated by combining this with the liquid oxygen density. Based on the multi-level threshold early warning model and the current environmental parameters, the remaining liquid oxygen mass is identified, and the liquid oxygen replenishment early warning level is determined. The remaining liquid oxygen mass is measured according to a preset time window, and the mass difference between adjacent time periods is calculated. Statistical analysis of the mass difference between adjacent time intervals yielded the average liquid oxygen consumption rate. Calculate the remaining available time based on the remaining liquid oxygen mass and the average liquid oxygen consumption rate; If the effective liquid oxygen flow rate is greater than the average liquid oxygen consumption rate, then an exponentially weighted moving average is applied to the average liquid oxygen consumption rate to calculate the available liquid oxygen time. No, then the remaining available time is determined to be the liquid oxygen available time; Based on the tank number, the liquid oxygen replenishment warning level is associated with the liquid oxygen availability time, and a liquid oxygen replenishment warning work order is generated.

[0015] By adopting the above technical solution, based on the multi-level threshold early warning model and the exponential weighted moving average algorithm, and through liquid oxygen data packet parsing, remaining mass conversion and average consumption rate statistics, the available time of liquid oxygen is dynamically calculated and graded replenishment work orders are automatically generated, which significantly improves the accuracy of liquid oxygen storage tank early warning and the timeliness of replenishment scheduling.

[0016] Secondly, this application also provides a remote monitoring and gas supply early warning system for medical liquid oxygen storage tanks based on the Internet of Things, which adopts the following technical solution: A remote monitoring and gas supply early warning system for medical liquid oxygen storage tanks based on the Internet of Things (IoT), comprising the following methods for implementing remote monitoring and gas supply early warning of medical liquid oxygen storage tanks: The data acquisition module is used to monitor the operating status of the liquid oxygen storage tank, collect multi-dimensional liquid oxygen storage data, and construct and transmit liquid oxygen data packets. The model building module is used to parse historical liquid oxygen parameter packages, combine them with tank geometric parameters, calculate the remaining liquid oxygen mass, and generate a multi-level threshold early warning model. The early warning and identification module is used to detect liquid oxygen data packets based on a multi-level threshold early warning model, calculate the available time of liquid oxygen, and generate liquid oxygen replenishment early warning work orders. The supply scheduling module is used to schedule liquid oxygen transport tanks to supply liquid oxygen storage tanks to the target liquid oxygen limit based on liquid oxygen supply early warning work orders.

[0017] The target liquid oxygen limit is the three-level safety warning threshold corresponding to different specifications of tanks, and is dynamically adjusted in combination with the liquid oxygen tank storage environment.

[0018] By adopting the above technical solution, based on the multi-level threshold early warning model and the exponential weighted moving average algorithm, and through liquid oxygen data packet parsing, remaining mass conversion and average consumption rate statistics, the available time of liquid oxygen is dynamically calculated and graded replenishment work orders are automatically generated, which significantly improves the accuracy of liquid oxygen storage tank early warning and the timeliness of replenishment scheduling.

[0019] Thirdly, this application also provides an electronic device, comprising: One or more processors; Memory, used to store one or more programs; When one or more programs are executed by one or more processors, the one or more processors implement any of the methods in the above scheme.

[0020] Fourthly, this application also provides a storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to realize the IoT-based remote monitoring and gas supply early warning method for medical liquid oxygen storage tanks as described above.

[0021] In summary, the beneficial technical effects of this application are as follows: Based on a multi-level threshold early warning model and an exponentially weighted moving average algorithm, the system dynamically calculates the available time of liquid oxygen and automatically generates tiered replenishment work orders by parsing liquid oxygen data packets, converting remaining mass, and statistically analyzing the average consumption rate. This improves the accuracy of early warning for liquid oxygen storage tanks and the timeliness of replenishment scheduling. Based on the extraction of consumption and storage features using a dual-branch attention network, a remaining mass weight matrix is ​​generated. Through hierarchical mapping and iterative correction using a dynamic hierarchical early warning function, the accurate construction of a multi-level threshold early warning model for liquid oxygen storage tanks is achieved, improving the reliability and adaptability of remaining mass prediction. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the remote monitoring and gas supply early warning method for medical liquid oxygen storage tanks in this application; Figure 2 This is a flowchart illustrating the multi-level threshold early warning model in this application; Figure 3 This is a schematic diagram of the remote monitoring and gas supply early warning system for medical liquid oxygen storage tanks in this application. Detailed Implementation

[0023] The present application will be further described in detail below with reference to the accompanying drawings.

[0024] Reference Figure 1 This application discloses a method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things, comprising: S1: Monitor the operating status of liquid oxygen storage tanks, collect multi-dimensional liquid oxygen storage data, and construct and transmit liquid oxygen data packages; S2: Analyze the historical liquid oxygen parameter package, combine it with the tank's geometric parameters, calculate the remaining liquid oxygen mass, generate a liquid oxygen time series data curve, and produce a multi-level threshold early warning model; S3: Detect liquid oxygen data packets based on the multi-level threshold early warning model, calculate the available time of liquid oxygen, and generate a liquid oxygen replenishment early warning work order; S4: Based on the liquid oxygen replenishment early warning work order, dispatch the liquid oxygen transport tank to replenish the liquid oxygen storage tank to the target liquid oxygen limit.

[0025] In this embodiment, pressure sensors, level sensors, and flow meters are installed on the liquid oxygen storage tank to collect real-time physical data. The edge gateway reads this sensor data periodically (e.g., every 5 minutes), encapsulates it into JSON format data packets, and sends them to the server in the logistics monitoring center via a 4G / 5G network or the hospital's intranet.

[0026] After receiving the data packet, the server parses out the liquid level, pressure, and instantaneous flow rate. The system converts the liquid level into the remaining liquid oxygen mass based on the tank's geometric parameters and stores this data in a time-series database, forming historical data curves.

[0027] The system has built-in "low liquid level warning threshold" and "pressure anomaly threshold". When the remaining liquid oxygen mass is lower than the first threshold (e.g., 20%), the system automatically generates a "replenishment warning" work order and pushes a notification to the gas supplier and the equipment department of the institute via SMS or APP; when the pressure value exceeds the safety limit, it triggers an audible and visual alarm and automatically issues a command to open the pressure relief valve.

[0028] The system calculates the average consumption rate based on the flow data of the past 24 hours and predicts the remaining availability of liquid oxygen. If the predicted availability is less than the preset safe time (e.g., 48 hours), the system automatically highlights the tank in the management backend and recommends the best replenishment time window.

[0029] Preferably, step S1 includes: The operating status of the liquid oxygen storage tank is detected according to the preset collection cycle to obtain the initial liquid oxygen pressure value, initial storage liquid level value and initial liquid oxygen flow rate value. The initial liquid oxygen pressure value, initial storage liquid level value, and initial liquid oxygen flow rate value are filtered, denoised, and anomaly removed to obtain the effective liquid oxygen pressure value, effective liquid oxygen level value, and effective liquid oxygen flow rate value. Based on the data collection timestamp, the effective liquid oxygen pressure value, effective liquid oxygen level value, and effective liquid oxygen flow rate value are correlated and converted, and combined with the liquid oxygen storage tank number, a liquid oxygen data package is constructed. The liquid oxygen data packets are compressed and encrypted based on the network bandwidth to obtain encrypted liquid oxygen packets, which are then transmitted and stored in a time-series database.

[0030] In this embodiment, a pressure sensor, a liquid level sensor, and a flow meter are installed at designated locations on the liquid oxygen storage tank to ensure that each sensor has a good seal with the tank interface and that its range is matched.

[0031] Connect the edge gateway to each sensor and set a timed data collection cycle (e.g., once every 5 minutes). At the same time, configure the network parameters (4G / 5G or hospital intranet) to connect to the logistics monitoring center server.

[0032] The edge gateway reads the real-time values ​​of the pressure sensor, liquid level sensor and flow meter simultaneously according to a preset cycle, and records the corresponding timestamps.

[0033] Perform range checks on the collected raw data, remove obviously abnormal or out-of-range values, and perform smoothing filtering if necessary.

[0034] The valid pressure values, liquid level values, flow rate values, and acquisition timestamps are organized into key-value pairs to form structured data objects.

[0035] The structured data object is converted into a standard JSON string, and metadata such as device ID and tank number are added to build a complete liquid oxygen data package.

[0036] If network bandwidth is limited or security requirements are high, JSON data packets can be lightly compressed (such as Gzip) and encrypted.

[0037] The edge gateway uploads the encapsulated liquid oxygen data packets to the server in the logistics monitoring center via a 4G / 5G network or the hospital's intranet.

[0038] After the server receives the data packet, it returns an acknowledgment message. The edge gateway records the collection and transmission logs for this event. If the transmission fails, the data packet is placed in the retry queue.

[0039] The server parses the data packets and stores the multi-dimensional liquid oxygen storage data into a time-series database for subsequent monitoring, alarms, and trend analysis.

[0040] Preferably, step S2 includes: The historical liquid oxygen parameter package was analyzed and its integrity was checked, and the tank number, liquid oxygen level, tank pressure value and instantaneous flow rate value were screened. The time-series database is matched based on the tank number to extract the corresponding tank geometric parameters; The corresponding liquid oxygen volume conversion function is called according to the shape of the storage tank, and the initial liquid oxygen mass is obtained by combining the tank geometric parameters and the liquid oxygen level height. Based on the tank temperature and the standard density table of liquid oxygen, the initial liquid oxygen mass is corrected for density, and the remaining liquid oxygen mass is calculated. The remaining liquid oxygen mass is interpolated temporally based on the collection timestamp to construct a liquid oxygen time-series data curve; Based on the tank pressure and instantaneous flow rate, the liquid oxygen time series data curve is auxiliaryly derived to form a periodic consumption time series curve; The cycle consumption time series curve and liquid oxygen time series data curve are mapped, divided and feature-iterated to determine multi-level liquid oxygen remaining mass thresholds and generate a multi-level threshold early warning model. Multi-stage liquid oxygen residual mass thresholds include: If the remaining mass of liquid oxygen is within the first consumption range, a mandatory replenishment notification will be generated and marked as a first-level mandatory early warning threshold. If the remaining mass of liquid oxygen is within the second consumption range, an emergency replenishment notification will be generated and marked as a level 2 emergency warning threshold. If the remaining mass of liquid oxygen is within the third consumption range, a planned replenishment notification will be generated and marked as a level three safety warning threshold. The warning thresholds for each level are dynamically adjusted based on the periodic consumption rate to form a dynamic hierarchical warning function.

[0041] In this embodiment, after receiving the JSON data packet sent by the edge gateway, the logistics monitoring center server parses out the key fields: device ID, tank number, data collection timestamp, liquid level (m), pressure (MPa), instantaneous flow rate (L / min), etc. Simultaneously, it verifies data integrity; if any necessary fields are missing, the data is marked as invalid and discarded.

[0042] Based on the tank number, retrieve the geometric parameters of the tank from the database, including: tank shape (cylindrical, spherical, or horizontal), tank diameter D (m), length L (m), end cap type (elliptical, dished, etc.), total tank volume V_total (m³), and density conversion factor. For horizontal cylindrical tanks, it is also necessary to extract the correspondence table (or fitting formula) between the horizontal cross-sectional area of ​​the tank and the liquid level.

[0043] Apply the appropriate volume calculation formula based on the shape of the storage tank: Vertical cylindrical tank: Volume V = Cross-sectional area × Liquid level height = (π × (D / 2)²) × H; Horizontal cylindrical tanks: use piecewise integral formulas or empirical formulas (such as the correction function of V=L×[(πD² / 4)×(H / D)]), or perform interpolation calculations through a pre-calibrated level-volume reference table.

[0044] Spherical tank: Calculated according to the formula for the volume of a spherical cap.

[0045] Calculate the liquid oxygen volume V_current at the current liquid level (unit: m³).

[0046] If the density of liquid oxygen is significantly affected by temperature and pressure, the temperature value can be obtained from the data package (if a temperature sensor is installed) or estimated from historical trends. Based on the real gas equation for liquid oxygen or a standard density table (e.g., a density of 1141 kg / m³ at the standard boiling point of -183°C), the volume is corrected for density to obtain the mass of liquid oxygen. Mass M = V_current × ρ_actual, where ρ_actual is interpolated based on the current pressure and temperature or corrected using the ideal gas law.

[0047] Based on the corrected volume and density, the remaining liquid oxygen mass (unit: kg or ton) is calculated and stored in the time-series database along with the acquisition timestamp and tank number. Simultaneously, the instantaneous flow rate is recorded to verify whether the rate of mass change matches the cumulative value from the flow meter.

[0048] The system queries the time-series database for remaining mass data within a specified time range (e.g., the last 24 hours, 7 days, or 30 days) and plots a continuous curve with time on the horizontal axis and mass on the vertical axis. Linear interpolation can be used to connect discrete points, and auxiliary curves such as pressure and instantaneous flow rate can be overlaid. The system automatically calculates derived indicators such as daily average consumption rate and remaining available time.

[0049] Based on hospital oxygen safety regulations and historical consumption patterns, multiple levels of liquid oxygen remaining mass thresholds (unit: tons or available hours) are set: Attention threshold (Level 3 warning): When the remaining quality is less than the 48-hour usage limit, the system sends a "caution" prompt to the logistics personnel.

[0050] Warning threshold (Level 2 warning): When the remaining quality is less than the 24-hour usage, the system issues a "warning" and notifies the duty supervisor.

[0051] Emergency threshold (Level 1 warning): When the remaining quality is less than 8 hours' worth of usage, the system immediately initiates the "emergency replenishment" process and sends an alarm to management.

[0052] Thresholds can be dynamically adjusted: for example, thresholds for each level can be automatically calculated based on the average consumption rate over the past week (e.g., T_emergency = average daily consumption × 0.33 days).

[0053] Reference Figure 2 Preferably, the multi-level threshold early warning model includes: Based on the first feature extraction branch, feature decomposition is performed on the periodic consumption time series curve to extract daily average consumption feature values ​​and monthly average consumption feature values; Based on the second feature extraction branch, feature analysis is performed on the liquid oxygen time series data curve to obtain the mass time series feature value and the tank application frequency feature value; Based on the first iteration training branch combined with the local attention mechanism, the daily average consumption feature value, monthly average consumption feature value and mass time series feature value are trained several times to generate the liquid oxygen mass change feature matrix. Based on the second iteration training branch combined with the spatiotemporal attention mechanism, the quality time series feature value, the tank application frequency feature value and the environmental change feature value are trained several times to generate the liquid oxygen storage and call feature matrix. Based on the integrated training branch, the liquid oxygen mass change feature matrix and the liquid oxygen storage and retrieval feature matrix are fused and spliced ​​to generate the liquid oxygen remaining mass weight matrix. Based on the auxiliary mapping branch and the multi-level liquid oxygen remaining mass threshold, the liquid oxygen remaining mass weight matrix is ​​mapped and differentiated to obtain the hierarchical warning weight matrix. Based on the identification optimization branch, the hierarchical warning weight matrix is ​​called to identify the historical liquid oxygen parameters hierarchically, and the hierarchical accuracy is calculated by combining the warning level label. If the hierarchical accuracy is less than the preset accuracy threshold, the hierarchical warning weight matrix is ​​corrected based on the dynamic hierarchical warning function and environmental change factors, and then identified again until the target is met. If not, then time-series fitting is performed on the dynamic hierarchical early warning function and the hierarchical warning weight matrix to generate a multi-level threshold early warning model.

[0054] In this embodiment, in the construction of an intelligent early warning model for liquid oxygen storage tanks in a large medical center, the following steps are taken: First, daily liquid oxygen consumption time-series data from the past three years, as well as historical liquid oxygen parameters such as tank filling records and ambient temperature and humidity, are collected. Then, the daily average consumption feature value (e.g., 32L / day) and monthly average consumption feature value (e.g., 960L / month) are extracted from the periodic consumption time-series curve through a first feature extraction branch. Next, the quality time-series feature value (tank pressure decay rate, liquid oxygen purity decrease gradient) and tank application frequency feature value (average 2.5 fillings per week) are obtained through analysis of the liquid oxygen time-series data curve using a second feature extraction branch. Based on the first iterative training branch... A local attention mechanism (with a 7-day window) is introduced to perform 200 rounds of iterative training on daily average consumption, monthly average consumption, and quality time-series features, generating a liquid oxygen quality change feature matrix that focuses on short-term fluctuations and long-term trends. The second iterative training branch adopts a spatiotemporal attention mechanism (simultaneously capturing time dependence and spatial distribution), performing 300 rounds of training on quality time-series, tank application frequency, and environmental change feature values ​​(such as daily temperature range and atmospheric pressure fluctuations), generating a liquid oxygen storage and retrieval feature matrix that reflects operating habits and external interference. The comprehensive training branch merges and splices the two matrices into a 256-dimensional liquid oxygen remaining quality weight matrix.

[0055] The auxiliary mapping branch maps the weight matrix into a five-level warning weight matrix based on multi-level residual mass thresholds (≥80% for normal, 60%-80% for Level 3 warning, 40%-60% for Level 2 warning, 20%-40% for Level 1 warning, and <20% for emergency). The identification and optimization branch uses this matrix to perform hierarchical identification on historical liquid oxygen parameters (a total of 1000 samples) and compares them with manually labeled warning levels. The calculated hierarchical accuracy is 87%, which is lower than the preset accuracy threshold of 95%. Environmental change factors are introduced through a dynamic hierarchical warning function. The consumption rate increases by 1.2 times for every 10°C increase in temperature. The hierarchical warning weight matrix is ​​corrected by adjusting the softmax temperature parameter from 1.0 to 1.3. After re-identification, the accuracy reaches 96%. Finally, the dynamic hierarchical warning function is time-series fitted with the corrected hierarchical warning weight matrix (using exponential smoothing with a smoothing factor of 0.7) to generate a multi-level threshold warning model that can adapt to environmental changes. This allows for accurate alarms to be issued 48 hours before the remaining liquid oxygen is close to the threshold, effectively ensuring the safety of oxygen supply in the ICU and operating room.

[0056] Preferably, step S3 includes: The liquid oxygen data packet is parsed to obtain the effective liquid oxygen level value and the effective liquid oxygen flow rate value; The effective liquid oxygen level is converted based on the tank's geometric parameters to obtain the effective liquid oxygen volume, and the remaining liquid oxygen mass is calculated by combining this with the liquid oxygen density. Based on the multi-level threshold early warning model and the current environmental parameters, the remaining liquid oxygen mass is identified, and the liquid oxygen replenishment early warning level is determined. The remaining liquid oxygen mass is measured according to a preset time window, and the mass difference between adjacent time periods is calculated. Statistical analysis of the mass difference between adjacent time intervals yielded the average liquid oxygen consumption rate. Calculate the remaining available time based on the remaining liquid oxygen mass and the average liquid oxygen consumption rate; If the effective liquid oxygen flow rate is greater than the average liquid oxygen consumption rate, then an exponentially weighted moving average is applied to the average liquid oxygen consumption rate to calculate the available liquid oxygen time. No, then the remaining available time is determined to be the liquid oxygen available time; Based on the tank number, the liquid oxygen replenishment warning level is associated with the liquid oxygen availability time, and a liquid oxygen replenishment warning work order is generated.

[0057] In this embodiment, the server receives the latest liquid oxygen data packet (JSON format) from the edge gateway and parses out parameters such as tank number, collection timestamp, current liquid level, pressure value, and instantaneous flow rate. Simultaneously, it reads historical data on the remaining liquid oxygen mass of the tank over a recent period (e.g., the past 24 hours) from a time-series database.

[0058] Convert the current liquid level height and tank geometric parameters (diameter, length, shape, etc.) to the current liquid oxygen volume; then combine the liquid oxygen density under the current pressure (which can be interpolated from a standard table or corrected using real-time temperature and pressure) to calculate the current remaining liquid oxygen mass Mcurrent (unit: kg or ton).

[0059] Extract the remaining mass data at regular time intervals (e.g., 1 hour) within the past 24 hours (or other set time window, such as 48 hours) from the time series database, calculate the mass difference between adjacent time points to obtain the hourly consumption. Take the median or average of these hourly consumptions as the current average consumption rate ravg (kg / h). If historical data is insufficient (for example, for a new tank), the design rated flow or industry empirical value can be used as the initial rate.

[0060] Calculate the remaining available time based on the current remaining mass and the average consumption rate: Tavailable=Mcurrent / ravg; Convert the unit to hours (h). At the same time, short-term trend correction can be performed in combination with the instantaneous flow rate: if the instantaneous flow rate is significantly higher than the average rate, the exponentially weighted moving average (EWMA) is used to dynamically update the rate, making the available time prediction more sensitive.

[0061] Read the preset warning level thresholds (for example: level three attention threshold T3=48 hours, level two alert threshold T2=24 hours, level one emergency threshold T1=8 hours). Compare the calculated Tavailable with the thresholds: If Tavailable≤T1: trigger level one warning (emergency).

[0062] If T1<Tavailable≤T2: trigger level two warning (alert).

[0063] If T2<Tavailable≤T3: trigger level three warning (attention).

[0064] If Tavailable>T3: no warning is triggered.

[0065] When any warning level is reached, the system automatically generates a warning work order. The content of the work order includes: work order number (automatically generated, such as LOX_ALARM_202505280001), storage tank number and location, current remaining mass (kg), calculated available time (hours), warning level and corresponding color identifier (red / orange / yellow), suggested replenishment window: for example, "Please complete the replenishment within 4 hours", generation timestamp, disposal status (initially "to be processed") and relevant responsible person (matched according to preset rules).

[0066] The system tracks the timeliness of early warning responses weekly or monthly. If, after multiple consecutive early warnings, actual replenishment is completed before the available time runs out, with a significant margin remaining, the system can automatically raise the thresholds at each level (e.g., by 5%) to reduce unnecessary early warnings. Conversely, if replenishment is frequently delayed until after an emergency warning, the thresholds will be lowered to issue warnings earlier. Adjustments must be approved by the administrator to take effect.

[0067] Preferably, step S4 includes: The system receives liquid oxygen replenishment alerts and extracts key information such as tank number, current location, current remaining liquid oxygen mass, available time, and alert level. Based on the alert level (Level 1 / Level 2 / Level 3), the system automatically determines the urgency of the replenishment (e.g., Level 1 requires departure within 1 hour, Level 3 can be scheduled within 4 hours).

[0068] Read the specifications of the storage tank (tank shape, diameter, length, total volume). The preset three-level safety warning thresholds correspond to different liquid level heights (e.g., Level 1 warning corresponds to 10% of the total height, Level 2 to 20%, and Level 3 to 30%). However, during actual replenishment, the target liquid oxygen limit is usually set to a safety upper limit higher than the Level 3 warning threshold, such as the Level 3 warning threshold plus a safety margin (e.g., 5% of the total volume). Simultaneously, dynamic adjustments are made based on the storage environment of the liquid oxygen tank. If the ambient temperature remains above 30°C (which accelerates the evaporation of liquid oxygen), the target liquid level will be increased by 2%.

[0069] If the average consumption rate over the past week is 10% higher than the historical average, the target liquid level will be increased by 3%.

[0070] If the storage tank is located outdoors and high temperatures are expected, increase the target liquid level in advance.

[0071] The final calculation yielded the target liquid oxygen mass Mtarget (unit: kg) and the corresponding target liquid level height Htarget for this replenishment.

[0072] Query the inventory of available liquid oxygen transport tanks (including location, current liquid oxygen balance, and tank number). Automatically select the most suitable transport tank based on urgency and distance (e.g., select the closest tank with sufficient liquid oxygen). Generate a dispatch instruction including: target tank location, target liquid oxygen quantity, required departure time, and estimated arrival time.

[0073] Once the transport tank arrives at the site, the operator uses a mobile terminal to scan the target storage tank's QR code (or NFC), and the system verifies a match in the task order. Then, following safety procedures, the transport tank's output hose is connected to the storage tank's input valve, ensuring a reliable grounding connection. After connection is complete, the operator confirms "Connection Ready" on the mobile terminal.

[0074] The operator clicks "Start Filling" on their mobile terminal, and the system opens the valve actuator, initiating liquid oxygen delivery. During the filling process, the system reads the tank's level sensor data once per second and displays the real-time difference between the current and target levels. Simultaneously, it monitors: Is the rate of increase in liquid level within the normal range (e.g., the increase does not exceed 2% of the total height per minute)?

[0075] Check if the pipeline pressure is abnormal. If an abnormality occurs (such as a sudden drop in the rate of increase), the system will automatically pause the refueling and prompt for inspection.

[0076] When the real-time liquid level reaches the target level Htarget, the system immediately sends a closing command to the valve actuator, automatically cutting off the liquid oxygen supply. The system records the refueling end time, final liquid level, and actual total refueling volume (calculated based on the liquid level difference and tank cross-sectional area). Simultaneously, the system generates a refueling completion report and updates the work order status to "Refueling Completed".

[0077] The system analyzes the response time, actual consumption rate, and deviation from predicted values ​​for each replenishment run weekly. If frequent emergency replenishments are detected due to insufficient environmental correction, the dynamic correction factor is automatically adjusted (e.g., increasing the temperature impact factor from 2% to 3%). Adjustments take effect only after administrator confirmation.

[0078] Reference Figure 3 A remote monitoring and gas supply early warning system for medical liquid oxygen storage tanks based on the Internet of Things, and a method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks, including: The data acquisition module is used to monitor the operating status of the liquid oxygen storage tank, collect multi-dimensional liquid oxygen storage data, and construct and transmit liquid oxygen data packets. The model building module is used to parse historical liquid oxygen parameter packages, combine them with tank geometric parameters, calculate the remaining liquid oxygen mass, and generate a multi-level threshold early warning model. The early warning and identification module is used to detect liquid oxygen data packets based on a multi-level threshold early warning model, calculate the available time of liquid oxygen, and generate liquid oxygen replenishment early warning work orders. The supply scheduling module is used to schedule liquid oxygen transport tanks to supply liquid oxygen storage tanks to the target liquid oxygen limit based on liquid oxygen supply early warning work orders.

[0079] The target liquid oxygen limit is the three-level safety warning threshold corresponding to different specifications of tanks, and is dynamically adjusted in combination with the liquid oxygen tank storage environment.

[0080] In this embodiment, the data acquisition module collects multi-dimensional liquid oxygen storage data, such as liquid level height, tank pressure, ambient temperature, and historical filling frequency, once per second, through high-frequency level gauges, pressure sensors, and temperature probes distributed on five storage tanks of different specifications. After compression and encryption by the edge gateway, the data is constructed into a timestamped liquid oxygen data packet and uploaded to the central monitoring system.

[0081] The model building module utilizes historical liquid oxygen parameter packages accumulated over the past two years, combined with the geometric parameters of each tank (such as diameter, height, and end cap shape), to accurately calculate the remaining liquid oxygen mass using the ideal gas law and mass transfer kinetics model. Based on tank type and dynamic consumption rate, it generates a multi-level threshold early warning model (Level 1 warning: remaining amount ≤15%, Level 2 warning: 15%-30%, Level 3 warning: 30%-45%, normal >45%). The early warning identification module monitors the incoming liquid oxygen data package in real time. When it calculates that a tank has only 18 hours of available time under the average daily consumption (standard is 2.5 tons / day) and the liquid level is below 17%, it automatically triggers a Level 2 warning and generates a liquid oxygen replenishment early warning work order containing the tank number, estimated depletion time, and suggested replenishment amount (e.g., 8 tons).

[0082] Based on the work order information, the supply scheduling module automatically dispatches the nearest liquid oxygen transport tanker to the site through GPS positioning and RFID identification. It also intelligently sets the target liquid level based on the tank storage environment (the target liquid oxygen limit is dynamically corrected to 90% during high summer temperatures and 95% during low winter temperatures). After the refueling is completed, the system automatically generates a closed-loop log and updates the multi-level threshold model parameters.

[0083] An electronic device, comprising: One or more processors; Memory, used to store one or more programs; When one or more programs are executed by one or more processors, the one or more processors implement any of the methods in the above scheme.

[0084] A storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to realize the remote monitoring and gas supply early warning method for medical liquid oxygen storage tanks as described above.

[0085] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things, characterized in that, include: Monitor the operating status of liquid oxygen storage tanks, collect multi-dimensional liquid oxygen storage data, and construct and transmit liquid oxygen data packages; Analyze historical liquid oxygen parameter packages, combine them with tank geometric parameters, calculate the remaining liquid oxygen mass, generate liquid oxygen time series data curves, and produce a multi-level threshold early warning model; The liquid oxygen data packet is detected based on the multi-level threshold early warning model, the available time of liquid oxygen is calculated, and a liquid oxygen replenishment early warning work order is generated. Based on the liquid oxygen replenishment early warning work order, the liquid oxygen transport tank is dispatched to replenish the liquid oxygen storage tank to the target liquid oxygen limit.

2. The method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things as described in claim 1, characterized in that, The monitoring of the liquid oxygen storage tank's operating status, the collection of multi-dimensional liquid oxygen storage data, and the construction and transmission of liquid oxygen data packets include: The operating status of the liquid oxygen storage tank is detected according to the preset collection cycle to obtain the initial liquid oxygen pressure value, initial storage liquid level value and initial liquid oxygen flow rate value. The initial liquid oxygen pressure value, initial storage liquid level value, and initial liquid oxygen flow rate value are filtered, denoised, and anomaly removed to obtain the effective liquid oxygen pressure value, effective liquid oxygen level value, and effective liquid oxygen flow rate value. Based on the data collection timestamp, the effective liquid oxygen pressure value, effective liquid oxygen level value, and effective liquid oxygen flow rate value are correlated and converted, and combined with the liquid oxygen storage tank number, a liquid oxygen data package is constructed. The liquid oxygen data packets are compressed and encrypted according to the network bandwidth to obtain encrypted liquid oxygen packets, which are then transmitted and stored in a time-series database.

3. The method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things according to claim 1, characterized in that, The analyzed historical liquid oxygen parameter package, combined with the tank's geometric parameters, calculates the remaining liquid oxygen mass, generates a liquid oxygen time-series data curve, and produces a multi-level threshold early warning model, including: The historical liquid oxygen parameter package was analyzed and its integrity was checked, and the tank number, liquid oxygen level, tank pressure value and instantaneous flow rate value were screened. The time-series database is matched based on the tank number to extract the corresponding tank geometric parameters; The corresponding liquid oxygen volume conversion function is called according to the shape of the storage tank, and the initial liquid oxygen mass is obtained by combining the tank geometric parameters and the liquid oxygen level height. Based on the tank temperature and the standard density table of liquid oxygen, the initial liquid oxygen mass is corrected for density, and the remaining liquid oxygen mass is calculated. The remaining liquid oxygen mass is interpolated temporally based on the collection timestamp to construct a liquid oxygen time-series data curve; Based on the tank pressure and instantaneous flow rate, the liquid oxygen time series data curve is further derived to form a periodic consumption time series curve. The cycle consumption time series curve and liquid oxygen time series data curve are mapped, divided, and feature iterated to determine the multi-level liquid oxygen remaining mass threshold and generate a multi-level threshold early warning model.

4. The method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things according to claim 3, characterized in that, The multi-stage liquid oxygen residual mass threshold includes: If the remaining mass of liquid oxygen is within the first consumption range, a mandatory replenishment notification will be generated and marked as a first-level mandatory early warning threshold. If the remaining mass of liquid oxygen is within the second consumption range, an emergency replenishment notification will be generated and marked as a level 2 emergency warning threshold. If the remaining mass of liquid oxygen is within the third consumption range, a planned replenishment notification will be generated and marked as a level three safety warning threshold. The warning thresholds for each level are dynamically adjusted based on the periodic consumption rate to form a dynamic hierarchical warning function.

5. The method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things according to claim 1, characterized in that, The multi-level threshold early warning model includes: Based on the first feature extraction branch, feature decomposition is performed on the periodic consumption time series curve to extract daily average consumption feature values ​​and monthly average consumption feature values; Based on the second feature extraction branch, feature analysis is performed on the liquid oxygen time series data curve to obtain the mass time series feature value and the tank application frequency feature value; Based on the first iteration training branch combined with the local attention mechanism, the daily average consumption feature value, monthly average consumption feature value and mass time series feature value are trained several times to generate the liquid oxygen mass change feature matrix. Based on the second iteration training branch combined with the spatiotemporal attention mechanism, the quality time series feature value, the tank application frequency feature value and the environmental change feature value are trained several times to generate the liquid oxygen storage and call feature matrix. Based on the integrated training branch, the liquid oxygen mass change feature matrix and the liquid oxygen storage and retrieval feature matrix are fused and spliced ​​to generate the liquid oxygen remaining mass weight matrix. Based on the auxiliary mapping branch and the multi-level liquid oxygen remaining mass threshold, the liquid oxygen remaining mass weight matrix is ​​mapped and differentiated to obtain the hierarchical warning weight matrix. Based on the identification optimization branch, the hierarchical warning weight matrix is ​​called to identify the historical liquid oxygen parameters hierarchically, and the hierarchical accuracy is calculated by combining the warning level label. If the accuracy of the level is less than the preset accuracy threshold, the level warning weight matrix is ​​corrected according to the dynamic hierarchical early warning function combined with environmental change factors, and the identification is repeated until the target is met. If not, then time-series fitting is performed on the dynamic hierarchical early warning function and the hierarchical warning weight matrix to generate a multi-level threshold early warning model.

6. The method for remote monitoring and gas supply early warning of medical liquid oxygen storage tanks based on the Internet of Things according to claim 1, characterized in that, The step of detecting the liquid oxygen data packet based on the multi-level threshold early warning model, calculating the available liquid oxygen time, and generating a liquid oxygen replenishment early warning work order includes: The liquid oxygen data packet is parsed to obtain the effective liquid oxygen level value and the effective liquid oxygen flow rate value; The effective liquid oxygen level is converted based on the tank's geometric parameters to obtain the effective liquid oxygen volume, and the remaining liquid oxygen mass is calculated by combining this with the liquid oxygen density. Based on the multi-level threshold early warning model and the current environmental parameters, the remaining liquid oxygen mass is identified, and the liquid oxygen replenishment early warning level is determined. The mass of the remaining liquid oxygen is measured according to a preset time window, and the mass difference between adjacent time periods is calculated. Statistical analysis of the mass difference between adjacent time intervals yielded the average liquid oxygen consumption rate. Calculate the remaining available time based on the remaining liquid oxygen mass and the average liquid oxygen consumption rate; If the effective liquid oxygen flow rate is greater than the average liquid oxygen consumption rate, then an exponentially weighted moving average is applied to the average liquid oxygen consumption rate to calculate the available liquid oxygen time. No, then the remaining available time is determined to be the liquid oxygen available time; Based on the tank number, the liquid oxygen replenishment warning level is associated with the liquid oxygen availability time, and a liquid oxygen replenishment warning work order is generated.

7. A remote monitoring and gas supply early warning system for medical liquid oxygen storage tanks based on the Internet of Things, used to implement the remote monitoring and gas supply early warning method for medical liquid oxygen storage tanks as described in any one of claims 1-6, characterized in that, include: The data acquisition module is used to monitor the operating status of the liquid oxygen storage tank, collect multi-dimensional liquid oxygen storage data, and construct and transmit liquid oxygen data packets. The model building module is used to parse historical liquid oxygen parameter packages, combine them with tank geometric parameters, calculate the remaining liquid oxygen mass, and generate a multi-level threshold early warning model. The early warning and identification module is used to detect the liquid oxygen data packet according to the multi-level threshold early warning model, calculate the liquid oxygen availability time, and generate a liquid oxygen replenishment early warning work order. The supply scheduling module is used to schedule liquid oxygen transport tanks to supply liquid oxygen storage tanks to the target liquid oxygen limit based on liquid oxygen supply early warning work orders.

8. The IoT-based remote monitoring and gas supply early warning system for medical liquid oxygen storage tanks according to claim 7, characterized in that: The target liquid oxygen limit is a three-level safety warning threshold corresponding to different specifications of tanks, and is dynamically adjusted in combination with the liquid oxygen tank storage environment.

9. A storage medium storing at least one instruction, at least one program, a code set, or an instruction set, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the remote monitoring and gas supply early warning method for a medical liquid oxygen storage tank as described in any one of claims 1 to 6.

Images