Medical gas detection method and device and storage medium

Abstract

The invention relates to the technical field of Internet of Things, in particular to a medical gas detection method and device and a storage medium, and the method comprises the following steps: collecting and processing multi-source data of a medical gas system in real time; characteristic factors are determined based on the multi-source data of the medical gas system, and system characteristic vectors are constructed; analyzing a pressure change rate, a first change rate parameter and a second change rate parameter according to the multi-source data; generating a change rate sequence according to the pressure change rate to extract a dominant frequency and judge a pressure change period, and marking an anomaly type based on the pressure change period, the pressure change rate, the first change rate parameter, the second change rate parameter and the system feature vector; training a random forest model based on the multi-source data and the period type; and using the trained random forest model to identify an abnormal state and perform early warning. According to the invention, accurate detection and early warning of medical gas are realized.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, and in particular to a medical gas detection method, device, and storage medium. Background Technology

[0002] Medical gas systems (such as oxygen and nitrogen) are critical hospital infrastructure, and their pressure stability directly impacts patient safety. Existing medical gas pressure monitoring systems typically employ fixed threshold alarm mechanisms, triggering an alarm when the pressure falls below a certain threshold. However, this mechanism has significant technical limitations: First, fixed thresholds cannot distinguish between temporary pressure fluctuations (such as brief peak gas usage) and genuine anomalies, leading to a high false alarm rate; second, fixed thresholds struggle to identify potential anomalies within pressure change trends (such as slow leaks), posing a risk of missed alarms; finally, existing systems rely on a single alarm strategy, lacking a coordinated processing mechanism for different anomaly types, hindering rapid and effective responses to emergencies. These problems severely impact the efficiency and safety of hospital logistics management.

[0003] Medical gas systems (such as oxygen and nitrogen) are critical hospital infrastructure, and their pressure stability directly impacts patient safety. Existing medical gas pressure monitoring systems typically employ fixed threshold alarm mechanisms, triggering an alarm when the pressure falls below a certain threshold. However, this mechanism suffers from the following technical drawbacks: high false alarm rate: temporary pressure fluctuations (such as brief peak gas usage) may trigger unnecessary alarms, leading to resource waste and management inefficiency; risk of missed alarms: fixed thresholds cannot identify potential anomalies (such as slow leaks) in pressure change trends, potentially causing serious safety hazards; limited response: alarms only notify relevant personnel, lacking a coordinated handling mechanism for different anomaly types, hindering rapid and effective response to emergencies. Summary of the Invention

[0004] The purpose of this invention is to provide a medical gas detection method, device, and storage medium to solve at least one of the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for detecting medical gases, comprising:

[0007] Real-time acquisition and processing of multi-source data from medical gas systems;

[0008] Feature factors were determined based on multi-source data from medical gas systems, and system feature vectors were constructed.

[0009] Based on multi-source data analysis, pressure change rate, first change rate parameter, and second change rate parameter are analyzed.

[0010] A rate of change sequence is generated based on the rate of pressure change to extract the dominant frequency and determine the pressure change cycle. Anomaly types are then marked based on the pressure change cycle, the rate of pressure change, the first rate of change parameter, the second rate of change parameter, and the system feature vector.

[0011] Random forest models are trained based on multi-source data and periodicity types;

[0012] Use a trained random forest model to identify and issue warnings about abnormal states.

[0013] Furthermore, the processed system pressure is normalized using a normalization formula, and the normalized system pressure is used as a pressure characteristic factor; the ratio of equipment current to rated current is used as a current characteristic factor; the ratio of ambient temperature to preset temperature parameter is used as a temperature characteristic factor; and a system characteristic vector is constructed based on the characteristic factors.

[0014] Furthermore, based on the analysis of pressure change rate using multi-source data, the expression for pressure change rate is: R(t)=[P(t)-P(t-1) / 1]×100%, where R(t) represents the pressure change rate and P(t-1) represents the system pressure at the previous time point;

[0015] Based on the pressure change rate analysis, the first change rate parameter is expressed as follows: In the formula, K1 represents the first rate of change parameter, and n represents the window parameter;

[0016] Based on the pressure change rate analysis, the second rate of change parameter is expressed as: K2=[R(t+1)+R(t-1)-2×R(t)] / Δt 2 In the formula, K2 represents the second rate of change parameter, and Δt represents the node time length parameter.

[0017] Furthermore, the pressure change rate is arranged in ascending order of time node number to obtain a change rate sequence, and a fast Fourier transform is performed on the change rate sequence to obtain the dominant frequency and pressure change period.

[0018] Furthermore, based on the pressure change rate, the first change rate parameter, the second change rate parameter, the system characteristic vector, and the pressure change period, the anomaly type is labeled. If R(t) ≤ r1 and K1 < k1 and K2 < k2, the anomaly type is labeled as a sudden pressure drop; if r1 ≤ R(t) ≤ r2 and σ(R(t)) ≤ α1, the anomaly type is labeled as a gradual pressure drop; if R(t) ≥ r3 and A2(t) ≥ α2 and P(t) / Pe ≥ α2, the anomaly type is labeled as a sudden pressure rise; if σ(R(t)) / μ(R(t)) ≥ α3 and AT ≤ at, the anomaly type is labeled as a pressure fluctuation; if |R(t)| If r1 ≥ r4 and σ(A1(t)) = σ(A2(t)) = σ(A2(t)) = 0, then the abnormality type is marked as sensor failure. Here, r1 represents the first rate of change threshold, r2 represents the second rate of change threshold, r3 represents the third rate of change threshold, r4 represents the fourth rate of change threshold, k1 represents the first parameter threshold, k2 represents the second parameter threshold, α1 represents the first fluctuation threshold, α2 represents the second fluctuation threshold, α3 represents the third fluctuation threshold, AT represents the pressure change period, at represents the period threshold, σ() represents the standard deviation of the data in parentheses, and μ() represents the average value of the data in parentheses.

[0019] Furthermore, when the period type is sensor failure, the rate of change threshold used when marking the anomaly type for the M time points before the current time point is updated, the first rate of change threshold and the second rate of change threshold are increased and the third rate of change threshold is decreased, where M represents the update range parameter.

[0020] Furthermore, based on the analysis of system pressure and equipment current time-domain characteristics, the expression of the time-domain characteristics is: H(t)=cov(P(t),I(t)) / [σ(P(t))×σ(I(t))], where H(t) represents the time-domain characteristics. The system feature vector, pressure change rate, first change rate parameter, second change rate parameter and time-domain characteristics are used as training features. The training features are mapped one-to-one with the period type, and the random forest model is trained using the training features and period type.

[0021] Furthermore, abnormal conditions are monitored and early warnings are issued. If the abnormal condition is a sudden drop in pressure, a level one alarm is triggered, the valve is closed, and maintenance personnel are notified. If the abnormal condition is a gradual drop in pressure, a level two alarm is triggered, and the gas supply department is notified to check the gas supply. If the abnormal condition is a sudden rise in pressure, a level three alarm is triggered, the valve status is checked, and the operator is notified. If the abnormal condition is a pressure fluctuation, a level four alarm is triggered, the data is recorded, and subsequent changes are observed.

[0022] On the other hand, the present invention also provides a medical gas detection device, comprising:

[0023] The acquisition module is used to acquire and process multi-source data from the medical gas system in real time.

[0024] A module is built to determine feature factors based on multi-source data from medical gas systems and to construct system feature vectors.

[0025] The analysis module is used to analyze the pressure change rate, the first change rate parameter, and the second change rate parameter based on multi-source data.

[0026] The labeling module is used to generate a rate of change sequence based on the rate of change of pressure, to extract the main frequency and determine the pressure change cycle, and to label the anomaly type based on the pressure change cycle, the rate of change of pressure, the first rate of change parameter, the second rate of change parameter and the system feature vector.

[0027] The training module is used to train random forest models based on multi-source data and periodicity types;

[0028] The early warning module is used to identify abnormal states and issue early warnings using a trained random forest model.

[0029] On the other hand, the present invention also provides a storage medium characterized in that it stores instructions that, when run on a computer, cause the computer to perform the medical gas detection method as described in any of the preceding claims.

[0030] The beneficial effects of this invention are as follows: by using multi-source data fusion and three-level rate of change parameters, it achieves a leap from threshold judgment to trend prediction, thereby improving the accuracy of anomaly identification; by using FFT periodic detection and random forest models, it enhances the anti-interference capability for complex working conditions; by adaptively updating the threshold when the sensor fails, it avoids cascading false alarms; by continuously iterating and training the model, it adapts to long-term changes such as system aging; and by using a four-level linkage mechanism, it improves the efficiency of accident handling and reduces sudden failures through preventive maintenance work orders. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 This is a flowchart of the medical gas detection method in this embodiment.

[0033] Figure 2 This is a flowchart of the method for constructing the system feature vector in this embodiment.

[0034] Figure 3 This is a flowchart of the exception type marking method in this embodiment.

[0035] Figure 4This is a schematic diagram of the medical gas detection device in this embodiment. Detailed Implementation

[0036] To more clearly illustrate the present invention, the following description, in conjunction with preferred embodiments and accompanying drawings, further explains the invention. Similar components in the drawings are indicated by the same reference numerals. Those skilled in the art should understand that the specific description below is illustrative rather than restrictive and should not be construed as limiting the scope of protection of the present invention.

[0037] It should be noted that although the terms first, second, third, etc., may be used in the embodiments of this application for description, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, without departing from the scope of the embodiments of this application, first can also be referred to as second, and similarly, second can also be referred to as first.

[0038] Please see Figure 1 As shown, this is the medical gas detection method of this embodiment, including:

[0039] Step S1: Real-time acquisition and processing of multi-source data from the medical gas system. The multi-source data includes pressure data, equipment current, rated current, and ambient temperature. The pressure data includes system pressure and rated pressure. The multi-source data is acquired by sensors installed on the medical gas system.

[0040] Specifically, in step S1 of this embodiment, a sliding window filter is used to process system stress, with the window width set to 5 seconds, and the filtering formula is: In the formula, P(t) represents the system pressure after processing, p(t1) represents the system pressure before processing, t represents the time node number, t1 represents the data time parameter, and NC represents the window width. The time node number is defined as the number used to distinguish data collected at different time nodes.

[0041] Specifically, in step S1 of this embodiment, a comprehensive monitoring foundation is constructed by fusing multi-dimensional data such as pressure, current, and temperature, and a sliding window filter is used to effectively suppress transient interference, improve data reliability, and avoid misjudgment.

[0042] Please continue reading. Figure 1 As shown, the medical gas detection method further includes:

[0043] Step S2: Determine feature factors based on multi-source data of the medical gas system and construct a system feature vector. The feature factors include pressure feature factors, current feature factors and temperature feature factors.

[0044] Please see Figure 2 As shown, this is a method for constructing system feature vectors, including:

[0045] Step S21: Determine feature factors based on multi-source data from the medical gas system.

[0046] Specifically, in step S21 of this embodiment, the processed system pressure is normalized using a normalization formula, and the normalized system pressure is used as a pressure characteristic factor. The normalization formula is: A1(t)=[P(t)-P(t)] min ] / [P(t) max -P(t) min In the formula, A1(t) represents the pressure characteristic factor, and P(t) represents the pressure characteristic factor. min P(t) represents the minimum pressure in the processed system. max This represents the maximum value of the processed system pressure.

[0047] Specifically, in step S21 of this embodiment, the ratio of the equipment current to the rated current is used as the current characteristic factor, and the current characteristic factor is set as A2(t), A2(t)=I(t) / Ie, where A2(t) represents the current characteristic factor, I(t) represents the equipment current, and Ie represents the rated current.

[0048] Specifically, in step S21 of this embodiment, the ratio of ambient temperature to preset temperature parameter is used as a temperature characteristic factor, and the temperature characteristic factor is set as A3(t), A3(t)=T(t) / T1, where A3(t) represents the temperature characteristic factor, T(t) represents the ambient temperature, and T1 represents the preset temperature parameter.

[0049] Specifically, in this embodiment, the preset temperature parameter is set to 50. This embodiment does not impose specific limitations on the setting of the preset temperature parameter, which can be freely set by those skilled in the art. The setting of the preset temperature parameter is related to the temperature that can meet the normal operation of the medical gas system.

[0050] Please continue reading. Figure 2 As shown, the method for constructing the system feature vector further includes:

[0051] Step S22: Construct system feature vectors based on feature factors.

[0052] Specifically, in step S22 of this embodiment, a system feature vector is constructed based on feature factors, and the system feature vector is set as V, V=[A1(t),A2(t),A3(t)], where V represents the system feature vector.

[0053] Specifically, in step S2 of this embodiment, pressure characteristic factors are normalized to eliminate differences in equipment dimensions and adapt to systems of different specifications. Current characteristic factors are used to directly reflect abnormal equipment loads. Temperature characteristic factors are used to correlate with environmental interference to prevent false alarms caused by thermal expansion. Feature vectors are used to quantify the system state and provide structured input for subsequent analysis.

[0054] Please continue reading. Figure 1 As shown, the medical gas detection method further includes:

[0055] Step S3: Analyze the pressure change rate, the first change rate parameter, and the second change rate parameter based on multi-source data.

[0056] Specifically, in step S3 of this embodiment, the pressure change rate is analyzed based on multi-source data. The expression for the pressure change rate is: R(t)=[P(t)-P(t-1) / 1]×100%, where R(t) represents the pressure change rate and P(t-1) represents the system pressure at the previous time point.

[0057] Specifically, in step S3 of this embodiment, a first rate of change parameter is analyzed based on the pressure change rate. The expression for the first rate of change parameter is: In the formula, K1 represents the first rate of change parameter, n represents the window parameter, and n∈{1,2,3,...,N}.

[0058] Specifically, in step S3 of this embodiment, a second rate of change parameter is analyzed based on the pressure change rate. The expression for the second rate of change parameter is: K2=[R(t+1)+R(t-1)-2×R(t)] / Δt 2 In the formula, K2 represents the second rate of change parameter, and Δt represents the node time length parameter. The node time length parameter is defined as the time interval between adjacent time nodes.

[0059] Specifically, in step S3 of this embodiment, sudden changes are captured in real time by using the pressure change rate, gradual trends are identified by using first-order parameters, inflection points are detected by using second-order parameters, and the entire scenario from instantaneous anomalies to progressive failures is covered by the complementary use of multiple parameters.

[0060] Please continue reading. Figure 1 As shown, the medical gas detection method further includes:

[0061] Step S4: Generate a rate of change sequence based on the pressure change rate to extract the dominant frequency and determine the pressure change cycle, and mark the anomaly type based on the pressure change cycle, pressure change rate, first rate of change parameter, second rate of change parameter and system feature vector.

[0062] Please see Figure 3As shown, it is a method for marking exception types, including:

[0063] Step S41: Generate a rate of change sequence based on the rate of pressure change to extract the dominant frequency and determine the pressure change cycle.

[0064] Specifically, in this embodiment, the pressure change rate is arranged sequentially in ascending order of time node number to obtain a change rate sequence. A Fast Fourier Transform is then performed on the change rate sequence to obtain the dominant frequency and the pressure change period. The pressure change period = 1 / dominant frequency, and the unit of the pressure change period is seconds.

[0065] Please continue reading. Figure 3 As shown, the method for marking the exception type further includes:

[0066] Step S42: Mark the anomaly type based on the pressure change rate, the first change rate parameter, the second change rate parameter, the system feature vector, and the pressure change cycle.

[0067] Specifically, in step S42 of this embodiment, the abnormality type is marked according to the pressure change rate, the first change rate parameter, the second change rate parameter, the system characteristic vector, and the pressure change period. If R(t)≤r1 and K1<k1 and K2<k2, the abnormality type is marked as a sudden pressure drop; if r1≤R(t)≤r2 and σ(R(t))≤α1, the abnormality type is marked as a gradual pressure drop; if R(t)≥r3 and A2(t)≥α2 and P(t) / Pe≥α2, the abnormality type is marked as a sudden pressure rise; if σ(R(t)) / μ(R(t))≥α3 and AT≤at, the abnormality type is marked as a pressure fluctuation; if |R(t)|≥r4 and σ(A1(t))=σ(A2(t))=σ(A2(t))=0, the abnormality type is marked as a sensor failure. Here, r1 represents the first... The threshold values ​​for the rate of change are: -12% ≤ r1 ≤ -10%, r2 represents the second threshold value for the rate of change, -5% ≤ r2 ≤ -1%, r3 represents the third threshold value for the rate of change, 6% ≤ r3 ≤ 8%, r4 represents the fourth threshold value for the rate of change, 5% ≤ r4 ≤ 8%, k1 represents the first parameter threshold value, -0.2 ≤ k1 ≤ -0.1%, k2 represents the second parameter threshold value, -0.12 ≤ k2 ≤ -0.08%, α1 represents the first fluctuation threshold value, 0.2 ≤ α1 ≤ 0.3%, α2 represents the second fluctuation threshold value, 1.2 ≤ α2 ≤ 1.3%, α3 represents the third fluctuation threshold value, 0.02 ≤ α3 ≤ 0.05%, AT represents the pressure change period, at represents the period threshold value, 150 ≤ at ≤ 200, σ() represents the standard deviation of the data in parentheses, μ() represents the average value of the data in parentheses, and Pe represents the rated pressure.

[0068] Specifically, in this embodiment, if any of the other situations mentioned above occur when marking the exception type, the exception type will not be marked.

[0069] Specifically, in step S42 of this embodiment, when the period type is sensor failure, the rate of change threshold used when marking the abnormal type for the M time nodes before the current time node is updated, the first rate of change threshold and the second rate of change threshold are increased and the third rate of change threshold is decreased, so that the first rate of change threshold and the second rate of change threshold are increased by X and the third rate of change threshold is decreased by X, where M represents the update range parameter, 100≤M≤500, and X represents the update change parameter, 1%≤X≤3%.

[0070] Specifically, in this embodiment, the update range parameter is set to 300 and the update change parameter is set to 2%. In this embodiment, the values ​​of the update range parameter and the update change parameter are not specifically limited, and those skilled in the art can set them freely.

[0071] Specifically, in this embodiment, the first rate of change threshold is set to -10%, the second rate of change threshold is set to -3%, the third rate of change threshold is set to 8%, the fourth rate of change threshold is set to 8%, the first parameter threshold is set to -0.15, the second parameter threshold is set to -0.1, the first fluctuation threshold is set to 0.3, the second fluctuation threshold is set to 1.3, the third fluctuation threshold is set to 0.05, and the period threshold is set to 180.

[0072] Specifically, in this embodiment, when judging the type of anomaly, if the system detects a rapid drop in pressure within a short period of time, it will be judged as a sudden pressure drop emergency fault. This situation is usually accompanied by a sudden increase in the variance of pressure fluctuations, and the curve shows a steep downward trend, which often indicates a major anomaly in the gas supply system. In actual operation, such anomalies are mostly caused by serious faults such as rupture of the main gas supply pipeline, accidental opening of critical valves, or malfunction of the safety valve of the gas storage tank. Once confirmed, the system will immediately activate the emergency response mechanism, including a series of measures such as cutting off the gas source in the affected area, activating the backup gas supply system, and triggering audible and visual alarm devices. At the same time, it will automatically push alarm information to the on-duty engineer's mobile APP and fully record the pressure data of the locked fault area for post-event analysis.

[0073] When the pressure value shows a continuous and stable downward trend, this type of abnormal pressure curve usually exhibits a smooth linear decline with a fluctuation variance generally less than 0.3, reflecting potential gradual losses or leaks in the system. In actual operation, such anomalies are often caused by minor leaks at pipe interfaces, performance degradation of pressure regulating valves, or abnormal gas consumption by gas-consuming equipment. Upon detecting such anomalies, the system immediately marks the abnormal area and activates a level-two alarm, while automatically generating a maintenance work order. To closely monitor the situation, the system automatically increases the monitoring frequency of this area from the usual once per minute to once every 10 seconds, and records detailed pressure change trend graphs to provide data support for subsequent troubleshooting. Although this type of anomaly will not immediately cause system failure, it may evolve into a more serious malfunction if not addressed promptly.

[0074] When an abnormally rapid rise in pressure is detected, this anomaly is often accompanied by abnormal fluctuations in equipment current, typically exceeding 30% of the normal value. This indicates a possible serious control failure in the gas supply system. In actual operation, such anomalies are often caused by pressure regulating valve failure, compressor output malfunction, or manual pressurization due to human error. Once the system confirms the anomaly, it will immediately activate the highest-level red alarm, first triggering the pressure relief device, simultaneously cutting off the upstream gas supply, and automatically notifying the technical supervisor to handle the situation. To facilitate accident analysis, the system will completely save all operational data for 10 minutes before and after the accident, including pressure curves, equipment current, valve status, and other key parameters, providing a complete basis for subsequent fault diagnosis and accountability.

[0075] When the monitored pressure values ​​exhibit regular, periodic fluctuations, this anomaly often reflects abnormal operating conditions of mechanical equipment. In actual operation, this is frequently caused by factors such as compressor piston wear, buffer tank failure, or loose pipeline fixation. Upon detecting such anomalies, the system immediately marks the malfunctioning equipment and generates a preventative maintenance work order. Simultaneously, it automatically adjusts equipment operating parameters to mitigate the abnormal situation. To aid in diagnosis, the system simultaneously initiates vibration monitoring, recording the fluctuation spectrum characteristics to provide a professional basis for subsequent targeted maintenance. Although such anomalies do not immediately affect the system's gas supply, their prolonged presence may lead to equipment damage or decreased energy efficiency, thus requiring timely intervention.

[0076] Specifically, in step S4 of this embodiment, the time-domain fluctuations are converted into frequency-domain periods by FFT main frequency extraction, the periodic fluctuations caused by mechanical vibration are accurately identified, and a five-dimensional judgment logic is set to reduce the false alarm rate and the false alarm rate.

[0077] Please continue reading. Figure 1 As shown, the medical gas detection method further includes:

[0078] Step S5: Train a random forest model based on multi-source data and periodicity type.

[0079] Specifically, in step S5 of this embodiment, the time-domain characteristics are analyzed based on system pressure and equipment current. The expression of the time-domain characteristics is: H(t)=cov(P(t),I(t)) / [σ(P(t))×σ(I(t))], where H(t) represents the time-domain characteristics. The system feature vector, pressure change rate, first change rate parameter, second change rate parameter, and time-domain characteristics are used as training features. The training features are mapped one-to-one with the period type. The random forest model is trained using the training features and period type. Each decision tree in the random forest model makes independent judgments, comparing feature values ​​layer by layer from the root, selecting paths according to pre-stored branching rules, and finally reaching the decision leaf node. The results of all decision trees are combined, and the number of votes for each anomaly type is counted. If the number of votes exceeds half, it is confirmed as the final result.

[0080] Specifically, in step S5 of this embodiment, by introducing time-domain features and associating electrical and pressure anomalies, the model accuracy is improved by fusing 6-dimensional features for training. The decision tree voting mechanism is used to avoid overfitting and has strong generalization ability.

[0081] Please continue reading. Figure 1 As shown, the medical gas detection method further includes:

[0082] Step S6: Use the trained random forest model to identify abnormal states and issue warnings.

[0083] Specifically, in step S6 of this embodiment, an abnormality warning is issued based on the abnormal state. If the abnormal state is a sudden drop in pressure, a level one alarm is triggered, the valve is closed, and maintenance personnel are notified. If the abnormal state is a gradual drop in pressure, a level two alarm is triggered, and the gas supply department is notified to check the gas supply. If the abnormal state is a sudden rise in pressure, a level three alarm is triggered, the valve status is checked, and the operator is notified. If the abnormal state is a pressure fluctuation, a level four alarm is triggered, the data is recorded, and subsequent changes are observed.

[0084] Please see Figure 4 As shown, this is the medical gas detection device of this embodiment, including:

[0085] The acquisition module is used to acquire and process multi-source data from the medical gas system in real time.

[0086] A module is built to determine feature factors based on multi-source data from medical gas systems and to construct system feature vectors.

[0087] The analysis module is used to analyze the pressure change rate, the first change rate parameter, and the second change rate parameter based on multi-source data.

[0088] The labeling module is used to generate a rate of change sequence based on the rate of change of pressure, to extract the main frequency and determine the pressure change cycle, and to label the anomaly type based on the pressure change cycle, the rate of change of pressure, the first rate of change parameter, the second rate of change parameter and the system feature vector.

[0089] The training module is used to train random forest models based on multi-source data and periodicity types;

[0090] The early warning module is used to identify abnormal states and issue early warnings using a trained random forest model.

[0091] This application also provides a computer-readable storage medium storing instructions that, when run on a computer, cause the computer to perform the medical gas detection method as described in the above method embodiments.

[0092] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable programs, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically contain computer-readable programs, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0093] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. For those skilled in the art, other variations or modifications can be made based on the above description. It is impossible to exhaustively list all the implementation methods here. All obvious variations or modifications derived from the technical solutions of the present invention are still within the protection scope of the present invention.

Claims

1. A method for detecting medical gases, characterized in that, include: Real-time acquisition and processing of multi-source data from medical gas systems; Feature factors were determined based on multi-source data from medical gas systems, and system feature vectors were constructed. Based on multi-source data analysis, pressure change rate, first change rate parameter, and second change rate parameter are analyzed. A rate of change sequence is generated based on the rate of pressure change to extract the dominant frequency and determine the pressure change cycle. Anomaly types are then marked based on the pressure change cycle, the rate of pressure change, the first rate of change parameter, the second rate of change parameter, and the system feature vector. Random forest models are trained based on multi-source data and periodicity types; Use a trained random forest model to identify and issue warnings about abnormal states.

2. The medical gas detection method according to claim 1, characterized in that, The processed system pressure is normalized using a normalization formula, and the normalized system pressure is used as a pressure characteristic factor. The ratio of the equipment current to the rated current is used as the current characteristic factor. The ratio of ambient temperature to preset temperature parameters is used as a temperature feature factor; and a system feature vector is constructed based on the feature factor.

3. The medical gas detection method according to claim 2, characterized in that, Based on multi-source data analysis of the pressure change rate, the expression for the pressure change rate is: R(t)=[P(t)-P(t-1) / 1]×100%, where R(t) represents the pressure change rate and P(t-1) represents the system pressure at the previous time point; Based on the pressure change rate analysis, the first change rate parameter is expressed as follows: In the formula, K1 represents the first rate of change parameter, and n represents the window parameter; Based on the pressure change rate analysis, the second rate of change parameter is expressed as: K2=[R(t+1)+R(t-1)-2×R(t)] / Δt 2 In the formula, K2 represents the second rate of change parameter, and Δt represents the node time length parameter.

4. The medical gas detection method according to claim 3, characterized in that, The pressure change rate is arranged in ascending order of time node number to obtain the change rate sequence. The change rate sequence is then subjected to a fast Fourier transform to obtain the dominant frequency and pressure change period.

5. The medical gas detection method according to claim 4, characterized in that, Anomaly types are identified based on the pressure change rate, the first change rate parameter, the second change rate parameter, the system characteristic vector, and the pressure change period. If R(t) ≤ r1 and K1 < k1 and K2 < k2, the anomaly type is identified as a sudden pressure drop; if r1 ≤ R(t) ≤ r2 and σ(R(t)) ≤ α1, the anomaly type is identified as a gradual pressure drop; if R(t) ≥ r3 and A2(t) ≥ α2 and P(t) / Pe ≥ α2, the anomaly type is identified as a sudden pressure rise; if σ(R(t)) / μ(R(t)) ≥ α3 and AT ≤ at, the anomaly type is identified as a pressure fluctuation; if |R(t)| ≥ r1, the anomaly type is identified as a pressure rise.

4. If σ(A1(t))=σ(A2(t))=σ(A2(t))=0, then the anomaly type is marked as sensor failure. Here, r1 represents the first rate of change threshold, r2 represents the second rate of change threshold, r3 represents the third rate of change threshold, r4 represents the fourth rate of change threshold, k1 represents the first parameter threshold, k2 represents the second parameter threshold, α1 represents the first fluctuation threshold, α2 represents the second fluctuation threshold, α3 represents the third fluctuation threshold, AT represents the pressure change period, at represents the period threshold, σ() represents the standard deviation of the data in parentheses, and μ() represents the average value of the data in parentheses.

6. The medical gas detection method according to claim 5, characterized in that, When the period type is sensor failure, update the rate of change threshold used when marking the anomaly type M time points before the current time point, increase the first rate of change threshold and the second rate of change threshold and decrease the third rate of change threshold, where M represents the update range parameter.

7. The medical gas detection method according to claim 6, characterized in that, Based on the analysis of system pressure and equipment current time-domain characteristics, the expression of the time-domain characteristics is: H(t)=cov(P(t),I(t)) / [σ(P(t))×σ(I(t))], where H(t) represents the time-domain characteristics. The system feature vector, pressure change rate, first change rate parameter, second change rate parameter and time-domain characteristics are used as training features. The training features are mapped one-to-one with the period type, and the random forest model is trained using the training features and period type.

8. The medical gas detection method according to claim 7, characterized in that, Anomaly warnings are issued based on abnormal conditions. If the abnormal condition is a sudden drop in pressure, a level 1 alarm is triggered, the valve is closed, and maintenance personnel are notified. If the abnormal condition is a gradual drop in pressure, a level 2 alarm is triggered, and the gas supply department is notified to check the gas supply. If the abnormal condition is a sudden rise in pressure, a level 3 alarm is triggered, the valve status is checked, and the operator is notified. If the abnormal condition is a pressure fluctuation, a level 4 alarm is triggered, the data is recorded, and subsequent changes are observed.

9. A medical gas detection device, applied to the medical gas detection method as described in any one of claims 1-8, characterized in that, include: The acquisition module is used to acquire and process multi-source data from the medical gas system in real time. A module is built to determine feature factors based on multi-source data from medical gas systems and to construct system feature vectors. The analysis module is used to analyze the pressure change rate, the first change rate parameter, and the second change rate parameter based on multi-source data. The labeling module is used to generate a rate of change sequence based on the rate of change of pressure, to extract the main frequency and determine the pressure change cycle, and to label the anomaly type based on the pressure change cycle, the rate of change of pressure, the first rate of change parameter, the second rate of change parameter and the system feature vector. The training module is used to train random forest models based on multi-source data and periodicity types; The early warning module is used to identify abnormal states and issue early warnings using a trained random forest model.

10. A storage medium, characterized in that, The device stores instructions that, when executed on a computer, cause the computer to perform the medical gas detection method as described in any one of claims 1-8.

Images