An edge communication control method for a medical oxygen generation system

Abstract

This invention relates to the field of machine-to-machine communication technology, specifically to an edge communication control method for a medical oxygen generation system. It tracks the inspiratory flow velocity trajectory to construct ventilation rhythm segments, calculates time offsets by combining chest cavity displacement responses, analyzes the offset evolution trend to screen for asynchronous cycles with increasing amplitude, synchronously monitors the flow velocity rise pattern to identify abrupt changes such as trend interruptions, configures oxygen supply actions for cycles exhibiting both asynchronous and abrupt characteristics, and integrates control commands to obtain an edge communication cycle control configuration set. This invention constructs rhythm intervals and multi-dimensional temporal mapping relationships using dual-channel data of inspiratory flow velocity and chest cavity displacement, tracks fluctuation evolution across cycles to doubly screen asynchronous responses and flow velocity abrupt changes, incorporates time offsets and synchronization states into the edge communication configuration by combining oxygen supply action insertion logic, and establishes a mechanism corresponding to rhythm numbers and status labels to achieve precise oxygen supply timing linkage matching and command control.

Description

Technical Field

[0001] This invention relates to the field of machine-to-machine communication technology, and more particularly to an edge communication control method for a medical oxygen generation system. Background Technology

[0002] Machine-to-machine (M2M) communication technology encompasses various technologies that enable automated information exchange and collaboration between devices. Its core content primarily involves achieving data synchronization, status monitoring, command transmission, and response control between terminal devices through wireless communication, wired transmission, and other methods. It is widely used in industries such as healthcare, power, transportation, and manufacturing, and is particularly prominent in smart devices, embedded systems, and edge computing scenarios. This field emphasizes real-time performance, stability, and data reliability between devices, playing a fundamental role in building automated and intelligent systems. From a systemic perspective, M2M communication technology covers multiple aspects, including communication protocol development, network architecture construction, data acquisition and uploading mechanisms, device identification and authentication mechanisms, communication fault diagnosis mechanisms, and multi-task coordination mechanisms between devices. It is an important component in building the Internet of Things (IoT) and intelligent control networks.

[0003] One of the edge communication control methods for a medical oxygen generation system refers to a specific method for achieving data transmission and collaborative control between the oxygen generation equipment and peripheral detection devices through a control mechanism deployed at the edge side. Addressing the need for linkage between the operating status of the oxygen generation equipment and the patient's physiological data in medical scenarios, this method utilizes an edge processing unit to receive, compare, and process parameters such as the patient's blood oxygen concentration and oxygen inhalation frequency, generating control commands and transmitting them to the oxygen generation device with low latency. This enables adaptive adjustment of the oxygen generation process. By establishing a communication channel with external monitoring devices, data exchange is completed using a short-range wireless communication protocol, and the communication content is parsed and responded to based on preset logic rules. This constructs a control mechanism for collaborative medical equipment based on edge computing.

[0004] In existing technologies, edge-side devices mainly generate control commands based on parameter thresholds. They lack a fine extraction mechanism for the fluctuation trend of ventilation actions and the start and end boundaries of rhythm within continuous cycles. When the patient's inspiratory rhythm is delayed, deviated, or abruptly changed, they cannot effectively identify it from the perspective of action continuity and rhythm synchronization. This can easily cause the device to misjudge the ventilation status in unstable cycles, resulting in a misalignment between the control response time and the actual physiological response. This reduces the continuous adaptation capability of the oxygen supply control rhythm. When facing non-steady-state problems such as abrupt changes in cycle rhythm and synchronization deviation, the handling strategy is simplistic and lacks dynamic identification and accurate response capabilities. Summary of the Invention

[0005] To address the technical problems existing in the prior art, embodiments of the present invention provide an edge communication control method for a medical oxygen generation system. The technical solution is as follows:

[0006] An edge communication control method for a medical oxygen generation system includes the following steps:

[0007] S1: Obtain the inspiratory flow rate sequence of the airflow channel in the medical oxygen generator, track the change of inspiratory flow rate from no ventilation to stable inspiratory flow, mark the continuous rise as the start point and the fall as the end point, define the inspiratory process interval, connect the sampling points to construct the continuous trajectory of inspiratory action, and generate ventilation rhythm segmentation.

[0008] S2: Based on the ventilation rhythm, divide the inspiratory start and end times in the segment, locate the segment, extract the thoracic displacement signal, identify the initial response time of the first continuous expansion, compare it with the inspiratory start time, determine the offset, summarize the comparison results of each cycle, and generate an expansion action time offset list.

[0009] S3: Analyze the continuity and fluctuation amplitude of the time offset of each rhythm cycle in the expansion action time offset list, track the offset evolution trend, filter the cycle segments with continuously increasing offset amplitude, extract the rhythm number to mark the asynchronous state, complete the state registration and perform cycle classification to obtain the asynchronous cycle identifier set.

[0010] S4: Extract the trajectory of inspiratory flow rate changes in the ventilation rhythm segment, analyze the flow rate change characteristics in the rising section, identify abrupt changes in trend interruption or reverse fluctuation, screen rhythmic cycles that meet the characteristics and annotate them, and generate a flow rate abrupt change cycle sequence.

[0011] As a further aspect of the present invention, the ventilation rhythm segmentation includes the inspiratory phase coverage area, the inspiratory flow rate time sequence trajectory, the respiratory action start and end markers, and the rhythm cycle number; the expansion action time offset list includes the initial response timestamp, rhythm correspondence, offset time interval, and cycle offset sequence; the asynchronous cycle identifier set includes the response asynchronous cycle number, offset amplitude change trend, cycle continuity label, and asynchronous state classification result; and the flow rate change cycle sequence includes the flow rate change cycle number, trend interruption type marker, rising segment change pattern, and change fluctuation amplitude.

[0012] As a further aspect of the present invention, the step of obtaining S1 is as follows:

[0013] S101: Obtain the inhalation flow rate sequence of the airflow channel in the medical oxygen generator, arrange all sampling point flow rate values ​​in chronological order, monitor the change difference between adjacent sampling points, screen for continuously rising segments, locate the position where the first continuous rising state occurs for more than two sampling points, take it as the starting point of the inhalation action, and generate an inhalation start position number.

[0014] S102: Based on the inhalation start position number, the inhalation flow rate sequence is called backward, the flow rate difference between the subsequent sampling point and the previous sampling point is judged, the position segment with a continuous descent state exceeding three sampling points is screened, the last sampling point is marked as the inhalation termination point, and the inhalation process coverage interval number is obtained.

[0015] S103: Based on the interval number covered by the inhalation process, extract the flow velocity values ​​of all sampling points in the corresponding segment, connect them in time order to construct the flow velocity trajectory, calculate the difference sequence between adjacent points, identify the fluctuation changes in the inhalation stage, integrate the data according to the original time series, and obtain the ventilation rhythm segmentation.

[0016] As a further aspect of the present invention, the step of obtaining S2 is as follows:

[0017] S201: Based on the inspiratory start and end times marked in the ventilation rhythm segment, locate the start and end positions of each inspiratory action in the time sequence, extract the thoracic surface displacement data of the corresponding time segment from the original displacement signal sequence, construct a continuous curve according to the sampling time order, and obtain the inspiratory segment displacement sequence.

[0018] S202: Based on the displacement sequence of the inhalation section, perform difference calculation on the displacement values ​​of adjacent sampling points, determine whether the continuous rising section is continuously greater than two sampling points, filter the earliest continuous expansion stage that meets the conditions, record the time position corresponding to the first rising point in the expansion stage, and obtain the expansion start response time point.

[0019] S203: Call the expansion start response time point, compare it with the start time of the corresponding inspiratory action in the ventilation rhythm segment, calculate the sampling point distance between the two, and combine it with the sampling frequency to convert it into a time interval to obtain the inspiratory expansion response offset duration;

[0020] S204: Based on the inspiratory expansion response offset duration, process the positional differences between the expansion start response time point and the inspiratory start time in all rhythmic cycles, extract the offset values ​​for each cycle, arrange them in order and unify the units to establish a set, and generate an expansion action time offset list.

[0021] As a further aspect of the present invention, the step of obtaining S3 is as follows:

[0022] S301: Based on the time offset data corresponding to each rhythm cycle in the expansion action time offset list, calculate the time offset difference sequence between adjacent cycles according to the temporal order of the rhythm cycle number, record the starting rhythm cycle number and offset difference value corresponding to each set of differences, determine the direction of change during the cycle by the sign of the change in the offset difference, screen the offset difference segments that continuously maintain a positive value, and generate the offset trend change segment.

[0023] S302: Call the cycle number and corresponding offset difference value recorded in the offset trend change segment, identify the cycle interval where the difference value range continuously increases, count the monotonically increasing degree of the offset difference in each interval, construct evaluation rules based on the number of cycles and the degree of increase, filter out the cycle number sequence that meets the increasing judgment condition, and generate a continuously enhanced cycle number sequence.

[0024] S303: Based on the continuously enhanced cycle numbering sequence, extract the time point information of each rhythm cycle before the ventilation command is triggered, record the continuous numbering relationship of the corresponding rhythm cycle in terms of offset performance, uniformly mark it as the asynchronous state of action response in time order, establish a mapping data table between rhythm cycle number and asynchronous mark, complete cycle classification and marking action, and generate an asynchronous cycle identifier set.

[0025] As a further aspect of the present invention, the step of obtaining S4 is as follows:

[0026] S401: Extract the trajectory of inspiratory flow rate change in each rhythm cycle of the ventilation rhythm segment, locate the segment in each cycle where the inspiratory flow rate rises from the beginning to the peak value, call the flow rate values ​​of all sampling points in the segment, construct the flow rate rise curve according to the sampling order, determine whether a monotonically increasing trend is formed based on the flow rate difference between consecutive sampling points, record the cycle number that does not meet the increasing condition separately, and generate an upward trend interruption number sequence.

[0027] S402: Based on the upward trend interruption number sequence, call the curve shape of the inhalation flow rate rising section within the corresponding period, and sequentially determine whether there is a situation where the flow rate value of any sampling point is less than the previous sampling point, or the flow rate values ​​of multiple sampling points remain approximately unchanged. Use the judgment standard that the flow rate difference is lower than the fluctuation threshold of the inhalation section to confirm the reverse fluctuation or stagnation characteristics, classify and label the period numbers that meet the conditions, and generate a list of abnormal inhalation morphology numbers.

[0028] S403: Call the rhythm cycle number in the list of abnormal inhalation patterns, obtain the flow rate difference between the start point of the inhalation flow rate of the corresponding rhythm cycle and the end point of the previous cycle, compare whether the difference is greater than the set inhalation mutation judgment threshold, filter the cycle numbers that meet the mutation conditions, construct the cycle sequence and attach the mutation status label, and generate the flow rate mutation cycle sequence.

[0029] As a further aspect of the present invention, the method further includes:

[0030] S5: Traverse the rhythm cycle numbers in the asynchronous cycle identifier set and compare them with the rhythm cycle numbers in the flow velocity mutation sequence, filter out rhythm cycles with dual characteristics, configure the oxygen supply sending action content, clarify the execution type, the stage to which it belongs and the synchronization identifier, integrate them into communication instructions, and generate an edge communication cycle control configuration set.

[0031] The edge communication cycle control configuration set includes oxygen supply action type, rhythm stage identifier, synchronization status parameters, control instruction content items, and communication cycle number.

[0032] As a further aspect of the present invention, the step of obtaining S5 is as follows:

[0033] S501: Traverse the rhythmic cycle numbers in the asynchronous cycle identifier set, call the ventilation rhythmic cycle corresponding to each number, obtain all rhythmic cycle numbers from the flow rate change cycle sequence, compare them one by one with the asynchronous numbers, confirm whether there is a double matching relationship in the number sequence, record the rhythmic cycle numbers that meet the double conditions and arrange them in the original sequence order to generate a joint feature cycle number set.

[0034] S502: Based on the joint feature cycle number set, obtain the ventilation rhythm segment corresponding to each number, extract the inspiratory flow rate change trajectory, rhythm stage information and synchronization status identifier in the segment, determine whether the current rhythm cycle number meets the starting conditions for dynamic adjustment of oxygen supply according to the position of the current rhythm cycle number in the cycle sequence, add oxygen supply action type, rhythm stage mark and synchronization status data to the cycle segment that meets the conditions, and generate oxygen supply action parameter set.

[0035] S503: Call the ventilation rhythm segment corresponding to each rhythm cycle in the oxygen supply action parameter set, insert the oxygen supply action parameters into the corresponding cycle segment structure in sequence, construct the edge control content item containing the oxygen supply action, integrate all cycle control units, arrange them in the rhythm sequence order to form a complete configuration dataset, and generate the edge communication cycle control configuration set.

[0036] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0037] In this invention, based on dual-channel data of inspiratory flow rate and thoracic displacement signal, a temporal mapping relationship of the start and end intervals of the ventilation rhythm, displacement response offset, and flow rate trend change is constructed by superimposing the data. By tracking the evolution path of fluctuation amplitude across cycles, a dual screening mechanism for the asynchronicity of action response and the change in flow rate is formed. Combined with the insertion method of oxygen supply action in rhythm segments, the response time offset, flow rate fluctuation pattern, and synchronization state parameters are incorporated into the communication configuration content to achieve linkage matching between rhythm control accuracy and oxygen supply timing adjustment. A correspondence between rhythm number and status label is established in the oxygen supply sending action configuration to ensure that the command execution has clear periodicity, targeted response, and status recognition capability. Attached Figure Description

[0038] Figure 1 This is a flowchart of the method of the present invention;

[0039] Figure 2 This is a flowchart illustrating the process of obtaining ventilation rhythm segmentation in this invention.

[0040] Figure 3 This is a flowchart illustrating the process of obtaining the expanded action time offset list in this invention.

[0041] Figure 4 This is a flowchart illustrating the process of obtaining the asynchronous periodic identifier set in this invention.

[0042] Figure 5 This is a flowchart illustrating the process of obtaining the flow velocity mutation cycle sequence of the present invention.

[0043] Figure 6 This is a flowchart illustrating the process of obtaining the edge communication cycle control configuration set according to the present invention. Detailed Implementation

[0044] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0045] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0046] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0047] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0048] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0049] Please see Figure 1 This invention provides a technical solution: an edge communication control method for a medical oxygen generation system, comprising the following steps:

[0050] S1: Obtain the inspiratory flow rate sequence of the airflow channel in the medical oxygen generator, track the change process of the inspiratory flow rate from the non-ventilated state to the stable inspiratory state in chronological order, mark the sampling position where the inspiratory flow rate first continuously rises as the inspiratory start point, continue to track the trend of inspiratory flow rate change along the inspiratory flow rate sequence, mark the sampling position where the inspiratory flow rate continuously falls and approaches the non-ventilated state as the inspiratory end point, take the inspiratory start point and end point as the boundary, continuously track the inspiratory flow rate fluctuation characteristics of the inspiratory stage, clearly define the complete coverage interval of the inspiratory process on the time axis, connect the sampling points in chronological order, construct the time continuous trajectory of the inspiratory action, and generate ventilation rhythm segmentation;

[0051] S2: Based on the inspiratory start and end times marked in the ventilation rhythm segment, locate the time segment corresponding to the inspiratory action, extract the raw data of the displacement signal on the thoracic surface, construct the time-series change curve, analyze the displacement trend, identify the first continuous expansion phase, determine the initial response time of the expansion action, extract the displacement signal timestamp corresponding to the initial response time, and compare the sequence position with the inspiratory start time marked in the ventilation rhythm segment to determine the time offset. Process all rhythm cycles in sequence, summarize the time offset results of each cycle, and generate a list of expansion action time offsets.

[0052] S3: Analyze the time offset of each rhythmic cycle in the expansion action time offset list, evaluate the continuity and fluctuation amplitude of the offset change cycle by cycle, track the offset evolution trend between cycles in the time dimension, detect whether there is a continuous increase in offset amplitude in multiple cycles, filter out the cycle segments with continuously increasing offset amplitude, extract the rhythmic cycle number corresponding to the cycle segment, mark it as the asynchronous state of action response, complete the state registration before the ventilation command is triggered for all marked asynchronous rhythmic cycles, and perform cycle classification based on the continuity of the response offset in the cycle sequence to obtain the asynchronous cycle identifier set;

[0053] S4: Extract the trajectory of inspiratory flow rate changes within each rhythmic cycle of the ventilation rhythm segment, locate the inspiratory flow rate rising segment, analyze the flow rate change characteristics during the rising process, determine whether there is a continuous rising trend, compare the change patterns of multiple rhythmic cycles, identify rhythmic cycles that show trend interruption, reverse fluctuation or stagnation during the rising process, determine whether the inspiratory flow rate exhibits abrupt behavior during adjacent ventilation rhythmic cycles, identify the abrupt state of inspiratory flow rate based on the abrupt behavior of the inspiratory flow rate change pattern, screen rhythmic cycles that meet the flow rate abrupt characteristics, annotate them as inspiratory abrupt cycle, and generate a flow rate abrupt cycle sequence.

[0054] S5: Traverse the rhythm cycle numbers in the asynchronous cycle identifier set, compare them one by one with the rhythm cycle numbers in the flow rate change cycle sequence, and select rhythm cycles that simultaneously possess asynchronous action response characteristics and inspiratory flow rate change characteristics. For rhythm cycles that meet the conditions for dynamic adjustment of oxygen supply, configure the corresponding oxygen supply sending action, insert the content item of the oxygen supply sending action into the matched ventilation rhythm segment, clarify the execution type, rhythm stage, and synchronization status identifier of the oxygen supply action, integrate all control content into communication control instructions, construct the edge communication control configuration unit of the current rhythm cycle, and generate the edge communication cycle control configuration set.

[0055] The ventilation rhythm segmentation includes the inspiratory phase coverage area, inspiratory flow rate time trajectory, respiratory action start and end markers, and rhythm cycle number. The expansion action time offset list includes the initial response timestamp, rhythm correspondence, offset time interval, and cycle offset sequence. The asynchronous cycle identifier set includes the response asynchronous cycle number, offset amplitude change trend, cycle continuity label, and asynchronous state classification result. The flow rate change cycle sequence includes the flow rate change cycle number, trend interruption type marker, rising segment change pattern, and change fluctuation amplitude. The edge communication cycle control configuration set includes the oxygen supply action type, rhythm phase identifier, synchronization state parameters, control instruction content items, and communication cycle number.

[0056] Please see Figure 2 The steps to obtain S1 are as follows:

[0057] S101: Obtain the inhalation flow rate sequence of the airflow channel in the medical oxygen generator, arrange all sampling point flow rate values ​​in chronological order, monitor the change difference between adjacent sampling points, screen for continuously rising segments, locate the position where the first continuous rising state occurs for more than two sampling points, take it as the starting point of the inhalation action, and generate an inhalation start position number.

[0058] To acquire the inhalation velocity sequence of the airflow channel in a medical oxygen generator, edge computing nodes read the analog voltage signal output from a MEMS thermal mass flow sensor via an analog-to-digital converter (ADC). The analog signal is discretized into digital velocity values ​​at a sampling frequency of 100Hz. The converted digital velocity values ​​are then written into a high-speed cache queue on the edge side, establishing a velocity sampling array of length N. ,in This represents the instantaneous flow velocity value at the i-th sampling point. The traversal pointer i is initially set to the first address of the array, and the current sampling point is read sequentially. With the next sampling point Perform subtraction on the given value. To obtain the flow velocity change difference between adjacent sampling points, a preset flow velocity noise fluctuation threshold is applied. The threshold is set based on the standard deviation of the flow rate sensor readings over 10 consecutive seconds when the device is in standby mode without ventilation. 3 times, that is For example, if the standard deviation of the flow rate reading measured in the standby test is 0.02 L / min, then set... The calculated difference is 0.06 L / min. and Perform numerical comparison, if If the current point is determined to be in an upward trend, the edge processing unit starts a temporary counter. And initialized to 0, when detected hour, Increment by 1;

[0059] If the next sampling point satisfies ,counter The value is incremented again to reach 2. At this point, it is determined that the condition of continuous upward movement exceeding two sampling points is met. The index of the position before the first meeting of the upward condition is locked back as the trigger starting index for the inhalation action. For example, in the flow rate sequence [0.05, 0.08, 0.16, 0.25, 0.40], the threshold is set to 0.06. 0.08-0.05=0.03 (<0.06, no count), 0.16-0.08=0.08 (>0.06, count 1), 0.25-0.16=0.09 (>0.06, count 2). Then the inhalation is determined to be valid, and the position corresponding to the value 0.08 at the starting point of the count is set. Write to the register to generate the intake start position number.

[0060] S102: Based on the inhalation start position number, the inhalation flow rate sequence is called backward, the flow rate difference between the subsequent sampling point and the previous sampling point is judged, the position segment with a continuous descent state exceeding three sampling points is screened, the last sampling point is marked as the inhalation termination point, and the inhalation process coverage interval number is obtained.

[0061] Based on the inhalation start position number, the inhalation flow rate sequence is invoked, and the edge computing unit reads the inhalation start position number from the register. traversal pointer Placed The position of the flow rate sampling array Read subsequent sampling point data along the index increment direction, and extract the sampling point at the current position respectively. Sampling point at the previous location The value is used to perform difference operations. Call the preset inhalation drop judgment threshold This threshold is set as the negative gradient reference for the peak inspiratory flow rate, with a value of -0.1 L / min. If the calculation result... If the point is determined to be in a declining state, the edge processing unit activates the continuous decline counter. Whenever a consecutive detection , Increment the value by 1, if it appears in the middle In this case, then Reset to zero, continue monitoring until... The value must be strictly greater than 3, meaning that four consecutive sampling points show a decreasing trend. For example, in the flow velocity sequence segment [25.0, 24.8, 24.2, 23.5, 22.0], the calculated difference sequence is [-0.2, -0.6, -0.7, -1.5]. If the threshold is set to -0.1, then all four differences are less than the threshold, satisfying the continuous decreasing condition. The edge node locks the position of the last sampling point that satisfies this condition. Mark this as the inflection point where the inspiratory flow rate returns to the baseline, and index this position. Defined as the inspiratory termination index , start index With Termination Index Combined packaging yields the numbering of the intake process coverage area.

[0062] S103: Based on the interval number covered by the inhalation process, extract the flow velocity values ​​of all sampling points in the corresponding segment, connect them in time order to construct the flow velocity trajectory, calculate the difference sequence between adjacent points, identify the fluctuation changes in the inhalation stage, integrate the data according to the original time series, and obtain the ventilation rhythm segmentation.

[0063] Based on the interval number covered by the inhalation process, the flow velocity values ​​of all sampling points within the corresponding segment are extracted, and the edge computing node parses the starting address contained in the interval number. and termination address From the full flow rate sampling array Perform a memory copy operation, copying the index range. Construct a subset of local suction air velocity based on all velocity data within the area. The data points in the subset are reconstructed in a virtual coordinate system according to the original sampling time sequence to form a discretized flow velocity trajectory vector. Then, adjacent element difference calculations are performed on this subset to generate a fluctuation feature sequence.

[0064] traversal sequence The absolute ratio of the sum of positive values ​​to the sum of negative values ​​in the sequence is calculated to quantify the stability of fluctuations during the inhalation phase. For example, for the extracted subset [0.5, 1.2, 2.0, 2.5, 2.3, 2.8, 3.0], the adjacent differences are calculated as [0.7, 0.8, 0.5, -0.2, 0.5, 0.2], identifying the non-monotonic fluctuation point (-0.2) within it, and then the velocity subset is... With fluctuation characteristic sequence The corresponding timestamp metadata is packaged to construct a data frame structure that conforms to the Edge Data Protocol. This data frame does not contain the original full data, but only the defined inspiratory payload, and obtains ventilation rhythm segmentation.

[0065] Please see Figure 3 The steps to obtain S2 are as follows:

[0066] S201: Based on the inspiratory start and end times marked in the ventilation rhythm segment, locate the start and end positions of each inspiratory action in the time sequence, extract the thoracic surface displacement data of the corresponding time segment from the original displacement signal sequence, construct a continuous curve according to the sampling time order, and obtain the inspiratory segment displacement sequence.

[0067] Based on the inspiratory start and end times calibrated in the ventilation rhythm segment, the edge computing node parses the ventilation rhythm segment data packet obtained from the flow rate sensor channel and extracts the start timestamp of each rhythm cycle. and end timestamp These two timestamps define the time window for the inhalation action. The edge nodes then access the original dataset of thoracic surface displacement signals stored in their local cache via the multimodal data bus. The dataset was collected in real time at a frequency of 50Hz by a piezoelectric respiratory belt worn on the user's chest and transmitted back to the edge gateway via Bluetooth. The edge processing unit then processed the original displacement dataset based on the parsed time window. Perform an index lookup operation to locate all results that meet the time condition. displacement sampling points ,in For the time stamp of the sampling point, For the corresponding displacement amplitude, all the located discrete sampling points are indexed according to time. Extract data in ascending order to a new memory buffer;

[0068] Construct an independent array structure ,in This represents the total number of sampling points within the inhalation window, assuming the start time of a certain inhalation. The termination time is 5000ms. If the sampling time is 6500ms and the sampling frequency is 50Hz, then the number of sampling points extracted is... Each edge node will have m consecutive displacement values. By connecting them sequentially, a numerical sequence reflecting the thoracic motion state during that specific inhalation period is formed, and the displacement sequence of the inhalation segment is obtained.

[0069] S202: Based on the displacement sequence of the inhalation section, perform difference calculation on the displacement values ​​of adjacent sampling points, determine whether the continuous rising section is continuously greater than two sampling points, filter the earliest continuous expansion stage that meets the conditions, record the time position corresponding to the first rising point in the expansion stage, and obtain the expansion start response time point.

[0070] Based on the displacement sequence of the intake section, the edge computing unit initializes a traversal pointer. Pointer sequence The first element sets up a status register to record the number of consecutive increases. The initial value is 0, and adjacent sampling points are read sequentially in the local processor. and The value;

[0071] Perform interpolation ;

[0072] Set a small displacement noise filtering threshold. This threshold is used to exclude the sensor's own electronic noise and slight body tremor interference. The calculated difference... If the current state is determined to be an expanded thoracic cavity, the edge node will register the state. If the value is increased by 1, Then immediately Reset to 0 and continue detecting the next set of data; the edge algorithm continuously monitors. The numerical change, once detected If three consecutive sampling points (corresponding to a duration of 60ms at 50Hz) show a valid upward trend, the edge node immediately stops traversing and locks the index of the first sampling point in the current consecutive upward sequence. For example, in the displacement sequence [10.0, 10.2, 10.8, 11.5, 12.3], 10.2 - 10.0 = 0.2 (< 0.5, reset), 10.8 - 10.2 = 0.6 (> 0.5, count 1), 11.5 - 10.8 = 0.7 (> 0.5, count 2), 12.3 - 11.5 = 0.8 (> 0.5, count 3). When traversing to 12.3, the continuously increasing count satisfies the condition, and the algorithm backtracks to lock the index position corresponding to the value 10.2, extracting that index position. The corresponding absolute timestamp on the original timeline This is marked as the earliest moment when the thoracic cavity produces a mechanical response to inhalation, thus obtaining the time point of the expansion initiation response.

[0073] S203: Call the expansion start response time point, compare it with the start time of the corresponding inspiratory action in the ventilation rhythm segment, calculate the sampling point distance between the two, and combine it with the sampling frequency to convert it into a time interval to obtain the inspiratory expansion response offset duration.

[0074] The edge computing node reads the locked expansion start response timestamp when the expansion start response time is invoked. Retrieve the inspiratory start timestamp corresponding to the current rhythm cycle from the ventilation rhythm data structure. Subtraction is performed in the arithmetic logic unit of the edge processor. This operation is performed directly based on the time index value of the underlying sampling points to obtain the difference in the number of sampling points between two events, and then calls the preset system sampling period parameter. This parameter is determined by the sensor configuration; for example, a 50Hz sampling rate corresponds to... Perform multiplication operations Convert the point difference to a standard time unit (milliseconds). For example, if the inspiratory initiation point index is 500 and the expansion response point index is 512, then the point difference is 12, and the time offset is calculated as follows: Edge nodes are calculated Numerical verification is performed. If the result is negative (i.e., expansion precedes airflow triggering), it is marked as an abnormally premature response. If the result exceeds a preset maximum physiological delay threshold (e.g., 500ms), it is marked as a hysteretic response. The verified values ​​are then... Numerical values ​​are retained as a key indicator characterizing human-machine synchronization performance within this period, and the inspiratory expansion response offset duration is obtained.

[0075] S204: Based on the inspiratory expansion response offset duration, process the positional difference between the expansion start response time point and the inspiratory start time in all rhythmic cycles, extract the offset values ​​under each cycle, arrange them in order and unify the units to establish a set, and generate an expansion action time offset list.

[0076] Based on the inspiratory-expansion response offset duration, the edge computing node initiates a loop processing process, traversing all ventilation rhythm segments stored in the local database, and performing processing for each rhythm cycle. Repeat the data extraction and calculation process from S201 to S203 to obtain the inhalation start time for each cycle. With expansion start response time ;

[0077] The offset duration of each cycle was calculated. The edge node allocates a dynamic list container in its local memory. , calculate The values ​​are sequentially filled into the container according to the order of occurrence of the rhythm cycle. For example, for three consecutive respiratory cycles, the offsets are calculated as [220ms, 240ms, 235ms], and these values ​​are stored in a list. Each value is appended with a corresponding cycle index ID. The data in the list is formatted to retain one decimal place and null values ​​(NaN) generated by invalid or erroneous calculations are removed. A structured dataset containing all historical cycle synchronization state data is constructed. This dataset resides in the non-volatile memory of the edge gateway and can be directly called by subsequent trend analysis algorithms to generate an expansion action time offset list.

[0078] Please see Figure 4 The steps to obtain S3 are as follows:

[0079] S301: Based on the time offset data corresponding to each rhythm cycle in the expansion action time offset list, calculate the difference sequence of time offset between adjacent cycles according to the temporal order of the rhythm cycle number, record the starting rhythm cycle number and offset difference value corresponding to each set of differences, determine the direction of change during the cycle by the sign of the change of the offset difference, screen the offset difference segments that continuously maintain positive values, and generate the offset trend change segment.

[0080] Based on the time offset data corresponding to each rhythm cycle in the expanded motion time offset list, the edge computing node loads the sorted expanded motion time offset list from its local cache. Initialize a difference calculation array Iterate through the elements in the list and read the first element. offset per cycle With the offset per cycle Perform subtraction operation The calculation results, along with the starting period number, will be used to calculate the results. In tuple form deposit Array, set the logic for determining positive and negative signs, if Mark the sign bit of that position as positive (Pos). Marked as negative (Neg), the edge node initiates a continuous scanning process. Search the array for a subsequence with a consecutive sign bit of Pos, and set a minimum consecutive length threshold. The value is 3, meaning that at least three consecutive periods must have a positive difference. For example, if the offset list is [200ms, 210ms, 225ms, 245ms, 240ms], the calculated difference is [+10, +15, +20, -5]. The first three differences must be consecutively positive and their length must meet the threshold. The edge node then assigns the period range corresponding to these three differences to an index. Lock, extract all within this range The tuple is used to construct an independent trend data block. Each data block is assigned a unique segment ID. These data blocks containing continuous positive drift characteristics are arranged in order of their position in the original sequence. Isolated positive points and segments with insufficient length are removed. Finally, all continuous positive difference segments that meet the conditions are integrated to generate the offset trend change segment.

[0081] S302: Call the cycle number and corresponding offset difference value recorded in the offset trend change segment, identify the cycle interval where the difference value range continuously increases, count the monotonically increasing degree of the offset difference in each interval, construct evaluation rules based on the number of cycles and the degree of increase, filter out the cycle number sequence that meets the increasing judgment condition, and generate a continuously enhanced cycle number sequence.

[0082] The edge computing unit retrieves the period number and corresponding offset difference value recorded in the offset trend change segment. For each extracted offset trend change segment, it performs internal gradient analysis, analyzing the difference sequence within the segment. Compare the magnitudes of adjacent differences item by item;

[0083] Execute the judgment logic: If If it is determined to be an accelerating increase, then it is considered to be increasing rapidly. If the speed is either constant or decelerating, a weighted scoring function is introduced for the edge nodes. ,in For step function (when The function value is 1 when the time condition is met, and 0 otherwise. Weighting coefficients that increase with sequence position (e.g.) This is used to amplify the weight of recent trends and set a scoring threshold. For example, for a difference sequence of length 4 [10,15,25,40], the differences increase progressively. When all values ​​in the function are 1, the accumulated weighted score exceeds the threshold, and the edge node indicates that the segment exhibits a "deteriorating" asynchronous trend, meaning the degree of human-machine asynchrony is accelerating. This applies to scores higher than [a certain threshold]. The segment is processed by extracting all the original rhythm cycle numbers contained therein. These numbers are arranged in ascending order to form an index array to be processed. Duplicate or overlapping numbers are removed. The edge nodes use this array to build an index view in memory, clearly identifying which specific respiratory cycles are in this abnormal state of accelerated shift, and generating a continuous enhanced cycle number sequence.

[0084] S303: Based on the continuous enhanced cycle number sequence, extract the time point information of each rhythm cycle before the ventilation command is triggered, record the continuous numbering relationship of the corresponding rhythm cycle in the offset performance, uniformly mark it as the asynchronous state of action response in time order, establish a mapping data table between rhythm cycle number and asynchronous mark, complete cycle classification and marking action, and generate an asynchronous cycle identifier set.

[0085] Based on the continuous enhancement cycle numbering sequence, the edge computing node accesses its local time scheduling table and retrieves each cycle number from the sequence. The corresponding next ventilation command preset trigger time ,exist Within the preceding time window (e.g., 50ms before triggering), the edge node sets a specific status flag in the device's status register for that period ID. Setting its value to 1 (True) indicates that the period is in an "asynchronous action response" alert state, and the edge node creates a hash mapping table. Using the rhythm cycle number as the key and a state object containing the asynchronous type (accelerated offset), current offset value, and predicted offset for the next cycle as the value, detailed state information for all marked cycles is written into this mapping table. For example, if cycle ID#105 is detected to have an accelerated offset, the current offset is 245ms, and the predicted offset for the next cycle may reach 270ms, then an entry is created in the mapping table.

[0086] {105:{Type:'Accel_Async',Curr:245,Pred:270}}, the edge node completes the registration of all selected cycles and locks the mapping table to a read-only state for subsequent control logic calls, ensuring that in the upcoming ventilation control cycle, the control algorithm can directly query the table to obtain a list of all cycles with synchronization problems and generate an asynchronous cycle identifier set.

[0087] Please see Figure 5 The steps to obtain S4 are as follows:

[0088] S401: Extract the trajectory of inspiratory flow rate changes in each rhythm cycle within the ventilation rhythm segment, locate the segment in each cycle where the inspiratory flow rate rises from the beginning to the peak value, call the flow rate values ​​of all sampling points in the segment, construct the flow rate rise curve according to the sampling order, determine whether a monotonically increasing trend is formed based on the flow rate difference between consecutive sampling points, record the cycle number that does not meet the increasing condition separately, and generate an upward trend interruption number sequence.

[0089] Extract the inspiratory flow rate variation trajectory within each rhythm cycle of the ventilation rhythm segment, and perform edge computing node traversal of the ventilation rhythm segment for each rhythm cycle. Analyze its flow rate data packets to locate the flow rate value that has reached its maximum value throughout the entire cycle. index position The starting index of this period to Extract all data points between them;

[0090] Constructing the velocity subsequence of the rising section ,in Corresponding to the initial flow velocity, Corresponding to the peak flow rate, the edge processing unit initializes a Boolean flag. Start the loop to compare the values ​​of adjacent sampling points one by one. and The judgment condition is If detected at any location This means that the flow velocity drops instantaneously, and the edge nodes immediately... Flag bit flipped to And terminate the scan of the current cycle. For example, in the subsequence [0.2, 0.5, 0.4, 0.8, 1.2], the value decreases from 0.5 to 0.4, which is determined to be non-monotonic. The edge nodes are all marked as... The period is determined, and its corresponding unique period identifier (UUID) is extracted and stored in a dedicated abnormal index array. This array exists as a temporary cache in the edge gateway memory and is used to collect all samples that are initially determined to be unstable in the upward trend, generating an upward trend interruption number sequence.

[0091] S402: Based on the interruption number sequence of the upward trend, call the curve shape of the inhalation flow rate rising section within the corresponding period, and sequentially determine whether there is a situation where the flow rate value of any sampling point is less than the previous sampling point, or the flow rate values ​​of multiple sampling points remain approximately unchanged. Use the judgment standard that the flow rate difference is lower than the fluctuation threshold of the inhalation section to confirm the reverse fluctuation or stagnation characteristics, classify and label the period numbers that meet the conditions, and generate a list of abnormal inhalation morphology numbers.

[0092] Based on the upward trend interruption number sequence, the edge computing unit reads the numbers one by one from the temporary buffer and backtracks to access the original flow rate increase segment data corresponding to each number. For each non-monotonic curve, morphological feature subdivision and discrimination are performed, and two micro-morphological thresholds are set:

[0093] Reverse fluctuation amplitude threshold (Indicates the minimum allowable measurement jitter limit) and the stagnation detection threshold (This represents the minimum resolution limit for changes in flow velocity).

[0094] First-order difference of the sequence calculated at edge nodes Traverse the difference array Detect whether it exists If present, the cycle is determined to have "reverse fluctuations" (i.e., obvious exhalation or airway closure during inhalation), and the presence of three or more consecutive difference values ​​is checked. If such a situation exists, it is determined that there is "flow stagnation" (i.e., insufficient inspiratory effort or airway obstruction) in that cycle, and the edge node constructs a state word for each abnormal cycle. Using bitmask technology, bit 0 is used to mark reverse fluctuations (0x01) and bit 1 is used to mark flow stagnation (0x02). For example, if a cycle has both reverse fluctuations and stagnation, its status word is 0x03. The cycle number is paired with the calculated status word and stored in a structured list to generate a list of abnormal intake morphology numbers.

[0095] S403: Call the rhythm cycle number in the list of abnormal inspiratory morphology numbers, obtain the flow rate difference between the start point of the inspiratory flow rate of the corresponding rhythm cycle and the end point of the previous cycle, compare whether the difference is greater than the set inspiratory mutation judgment threshold, filter the cycle numbers that meet the mutation conditions, construct the cycle sequence and attach mutation status label, and generate the flow rate mutation cycle sequence.

[0096] The edge computing node retrieves the rhythmic cycle number from the list of abnormal inspiratory morphology numbers, and then targets each abnormal cycle in the list. Extract the velocity value at the starting point of its suction flow. It also traces back to obtain the inspiratory end point flow rate value of the immediately preceding rhythm cycle. These two points represent the two ends of the apnea interval on the time axis, and the edge nodes perform cross-cycle interpolation:

[0097] Calculate the amplitude of the flow rate baseline jump at the moment of inspiratory action switching, and set the threshold for inspiratory abrupt change judgment. This threshold is dynamically correlated with the device's currently set base traffic. The value is For example, if the current base flow rate is 2L / min, then Edge node judgment if This indicates a significant airflow baseline shift or abrupt inspiratory interruption between two respiratory cycles. In this case, the cycle is classified as a "mutation cycle," and the edge node creates a control descriptor for this cycle, which includes the cycle index, mutation type (Baseline_Shift), and jump magnitude. In addition to the morphological anomaly bitmask identified in the preceding sequence, these descriptors are linked in chronological order of occurrence to construct a doubly linked list structure. This linked list is dedicated to storing flow velocity mutation events that require high-priority intervention, generating a flow velocity mutation cycle sequence.

[0098] Please see Figure 6 The steps to obtain S5 are as follows:

[0099] S501: Traverse the rhythmic cycle numbers in the asynchronous cycle identifier set, call the ventilation rhythmic cycle corresponding to each number, obtain all rhythmic cycle numbers from the flow velocity change cycle sequence, compare them one by one with the asynchronous numbers, confirm whether there is a double matching relationship in the number sequence, record the rhythmic cycle numbers that meet the double conditions and arrange them in the original sequence order to generate a joint feature cycle number set.

[0100] The edge computing node iterates through the rhythm cycle numbers in the asynchronous cycle identifier set and reads the local hash map. All key values ​​(Keys) represent the set of rhythmic cycle IDs marked as "Asynchronous Action Response" in step S3. The edge node accesses the doubly linked list generated in step S4, traverses the flow velocity mutation cycle sequences stored in it, extracts all cycle IDs with mutation state descriptors, and constructs another set. The edge processing unit uses the set intersection operation rule to perform logical operations. The selection process identifies elements that exist in both sets simultaneously, specifically those cycles that exhibit both asynchronous breathing and abrupt changes in inspiratory flow rate (instability), such as set [e.g., ...]]]]]]]]]]] The set containing {102, 105, 108, 112} Includes {105, 106, 112, 115};

[0101] The target set is obtained after the intersection operation. The edge nodes assign priority weights to the two matching cycle IDs. Given the potential oxygen supply risk from the superposition of these two anomalies, their priority is set to the highest level (Level_Critical). The edge nodes then classify the anomalies according to the time sequence number of the original rhythm cycle. The elements in the sequence are quickly sorted to ensure that the output sequence strictly follows the chronological order of occurrence, which facilitates the generation of control commands in a timely manner. The sorted ID list is written into the task scheduling queue of the edge gateway as the core object for the upcoming oxygen supply strategy adjustment, and a joint feature periodic number set is generated.

[0102] S502: Based on the joint feature cycle number set, obtain the ventilation rhythm segment corresponding to each number, extract the inspiratory flow rate change trajectory, rhythm stage information and synchronization status identifier in the segment, determine whether the current rhythm cycle number meets the starting conditions for dynamic adjustment of oxygen supply based on the position of the current rhythm cycle number in the cycle sequence, add oxygen supply action type, rhythm stage mark and synchronization status data to the cycle segment that meets the conditions, and generate oxygen supply action parameter set.

[0103] Based on the joint feature cycle number set, the edge computing node sequentially retrieves the target cycle ID from the task scheduling queue. For each ID, it uses an index pointer to backtrack and access the original ventilation rhythm segment data structure. It directly reads from memory the real-time inspiratory velocity trajectory array corresponding to that cycle, the current rhythm stage (e.g., early inspiratory phase, mid-inspiratory phase, late inspiratory phase), and the synchronization status flag (e.g., 200ms lag). The edge node introduces a dynamic oxygen supply adjustment decision logic: it checks whether the cycle ID is located in the preceding position of the current system clock (i.e., belonging to a historically occurring cycle) or within the current or future prediction window (i.e., belonging to a pending cycle). If it belongs to a pending cycle, it determines that the dynamic oxygen supply adjustment start condition is met. For cycles that meet the condition, the edge node configures specific oxygen supply action parameters: setting the "oxygen supply action type" to "Pulse Compensation Mode" (Pulse_Compensate), setting the "trigger delay" to 0ms (i.e., immediate triggering), and adjusting based on the offset calculated in step S2. Dynamic calculation of oxygen pulse width:

[0104] ,in Base pulse width (e.g., 50ms). The dimensions are Time compensation coefficient (e.g., 0.002) If the offset is 200ms, the pulse width is adjusted to... The edge node encapsulates the calculated pulse width, trigger mode, and associated rhythm phase markers (such as Phase_Inhale_Early) into a standardized JSON format data packet, which serves as the control parameter carrier for this specific abnormal cycle, generating the oxygen supply action parameter set.

[0105] S503: Call the ventilation rhythm segment corresponding to each rhythm cycle in the oxygen supply action parameter set, insert the oxygen supply action parameters into the corresponding cycle segment structure in sequence, construct the edge control content item containing the oxygen supply action, integrate all cycle control units, arrange them in the rhythm sequence order to form a complete configuration dataset, and generate the edge communication cycle control configuration set.

[0106] The edge computing node invokes the ventilation rhythm segments corresponding to each rhythm cycle in the oxygen supply action parameter set. It then starts the edge communication control signal generator. For each target cycle with generated oxygen supply action parameters, it creates a control message structure conforming to the MQTT or CoAP communication protocol standard. The Payload field of this message contains the basic timing information (start time, duration) of the ventilation rhythm segment, embedding the oxygen supply action parameter JSON object within it. This explicitly specifies the action command (Open_Valve, Duration: 70m) that the edge actuator (such as a solenoid valve drive circuit) should execute at a specific time point. (s) The edge node further integrates these independent control messages. For continuous control commands, a batch processing method is used to merge them to build a serialized data stream containing multiple control units. For example, control commands for cycles #105 and #112 are packaged into the same downlink control frame, and a CRC check code and the edge node's device authentication signature are added to the frame header to ensure the integrity and security of the commands during transmission. The resulting binary data block is a complete set of commands that can be directly sent to the underlying hardware actuators through the wireless module. This set of commands no longer relies on cloud confirmation and takes effect directly in a closed loop at the edge, generating the edge communication cycle control configuration set.

[0107] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for controlling edge communication of a medical oxygen generation system, characterized by, Includes the following steps: S1: Obtain the inspiratory flow rate sequence of the airflow channel in the medical oxygen generator, track the change of inspiratory flow rate from no ventilation to stable inspiratory flow, mark the continuous rise as the start point and the fall as the end point, define the inspiratory process interval, connect the sampling points to construct the continuous trajectory of inspiratory action, and generate ventilation rhythm segmentation. The steps for obtaining S1 are as follows: S101: Obtain the inhalation flow rate sequence of the airflow channel in the medical oxygen generator, arrange all sampling point flow rate values ​​in chronological order, monitor the change difference between adjacent sampling points, screen for continuously rising segments, locate the position where the first continuous rising state occurs for more than two sampling points, take it as the starting point of the inhalation action, and generate an inhalation start position number. S102: Based on the inhalation start position number, the inhalation flow rate sequence is called backward, the flow rate difference between the subsequent sampling point and the previous sampling point is judged, the position segment with a cumulative total of more than three sampling points in the continuous descent state is screened, the last sampling point is marked as the inhalation termination point, and the inhalation process coverage interval number is obtained. S103: Based on the interval number covered by the inhalation process, extract the flow velocity values ​​of all sampling points in the corresponding segment, connect them in time order to construct the flow velocity trajectory, calculate the difference sequence between adjacent points, identify the fluctuation changes in the inhalation stage, integrate the data according to the original time sequence, and obtain the ventilation rhythm segmentation. S2: Based on the ventilation rhythm, divide the inspiratory start and end times in the segment, locate the segment, extract the thoracic displacement signal, identify the initial response time of the first continuous expansion, compare it with the inspiratory start time, determine the offset, summarize the comparison results of each cycle, and generate an expansion action time offset list. The steps for obtaining S2 are as follows: S201: Based on the inspiratory start and end times marked in the ventilation rhythm segment, locate the start and end positions of each inspiratory action in the time sequence, extract the thoracic surface displacement data of the corresponding time segment from the original displacement signal sequence, construct a continuous curve according to the sampling time order, and obtain the inspiratory segment displacement sequence. S202: Based on the displacement sequence of the inhalation section, perform difference calculation on the displacement values ​​of adjacent sampling points, determine whether the continuous rising section is continuously greater than two sampling points, filter the earliest continuous expansion stage that meets the conditions, record the time position corresponding to the first rising point in the expansion stage, and obtain the expansion start response time point. S203: Call the expansion start response time point, compare it with the start time of the corresponding inspiratory action in the ventilation rhythm segment, calculate the sampling point distance between the two, and combine it with the sampling frequency to convert it into a time interval to obtain the inspiratory expansion response offset duration; S204: Based on the inspiratory expansion response offset duration, process the positional difference between the expansion start response time point and the inspiratory start time in all rhythmic cycles, extract the offset value under each cycle, arrange them in order and unify the units to establish a set, and generate an expansion action time offset list. S3: Analyze the continuity and fluctuation amplitude of the time offset of each rhythm cycle in the expansion action time offset list, track the offset evolution trend, filter the cycle segments with continuously increasing offset amplitude, extract the rhythm number to mark the asynchronous state, complete the state registration and perform cycle classification to obtain the asynchronous cycle identifier set. The steps for obtaining S3 are as follows: S301: Based on the time offset data corresponding to each rhythm cycle in the expansion action time offset list, calculate the time offset difference sequence between adjacent cycles according to the temporal order of the rhythm cycle number, record the starting rhythm cycle number and offset difference value corresponding to each set of differences, determine the direction of change during the cycle by the sign of the change in the offset difference, screen the offset difference segments that continuously maintain a positive value, and generate the offset trend change segment. S302: Call the cycle number and corresponding offset difference value recorded in the offset trend change segment, identify the cycle interval where the difference value range continuously increases, count the monotonically increasing degree of the offset difference in each interval, construct evaluation rules based on the number of cycles and the degree of increase, filter out the cycle number sequence that meets the increasing judgment condition, and generate a continuously enhanced cycle number sequence. S303: Based on the continuous enhanced cycle numbering sequence, extract the time point information of each rhythm cycle before the ventilation command is triggered, record the continuous numbering relationship of the corresponding rhythm cycle in the offset performance, uniformly mark it as the asynchronous state of action response in time order, establish a mapping data table between rhythm cycle number and asynchronous mark, complete cycle classification and marking action, and generate an asynchronous cycle identifier set. S4: Extract the inspiratory flow velocity change trajectory in the ventilation rhythm segment, analyze the flow velocity change characteristics in the rising section, identify abrupt changes in trend interruption or reverse fluctuation, screen out rhythmic cycles that meet the characteristics and annotate them, and generate a flow velocity abrupt change cycle sequence. The steps for obtaining S4 are as follows: S401: Extract the trajectory of inspiratory flow rate change in each rhythm cycle of the ventilation rhythm segment, locate the segment in each cycle where the inspiratory flow rate rises from the beginning to the peak value, call the flow rate values ​​of all sampling points in the segment, construct the flow rate rise curve according to the sampling order, determine whether a monotonically increasing trend is formed based on the flow rate difference between consecutive sampling points, record the cycle number that does not meet the increasing condition separately, and generate an upward trend interruption number sequence. S402: Based on the upward trend interruption number sequence, call the curve shape of the inhalation flow rate rising section within the corresponding period, and sequentially determine whether there is a situation where the flow rate value of any sampling point is less than the previous sampling point, or the flow rate values ​​of multiple sampling points remain approximately unchanged. Use the judgment standard that the flow rate difference is lower than the fluctuation threshold of the inhalation section to confirm the reverse fluctuation or stagnation characteristics, classify and label the period numbers that meet the conditions, and generate a list of abnormal inhalation morphology numbers. S403: Call the rhythm cycle number in the list of abnormal inspiratory morphology numbers, obtain the flow rate difference between the start point of the inspiratory flow rate of the corresponding rhythm cycle and the end point of the previous cycle, compare whether the difference is greater than the set inspiratory mutation judgment threshold, filter the cycle number that meets the mutation condition, construct the cycle sequence and attach the mutation status label, and generate the flow rate mutation cycle sequence. S5: Traverse the rhythm cycle numbers in the asynchronous cycle identifier set and compare them with the rhythm cycle numbers in the flow velocity mutation sequence, filter out rhythm cycles with dual characteristics, configure the oxygen supply sending action content, clarify the execution type, the stage to which it belongs and the synchronization identifier, integrate them into communication instructions, and generate an edge communication cycle control configuration set. The steps for obtaining S5 are as follows: S501: Traverse the rhythmic cycle numbers in the asynchronous cycle identifier set, call the ventilation rhythmic cycle corresponding to each number, obtain all rhythmic cycle numbers from the flow rate change cycle sequence, compare them one by one with the asynchronous numbers, confirm whether there is a double matching relationship in the number sequence, record the rhythmic cycle numbers that meet the double conditions and arrange them in the original sequence order to generate a joint feature cycle number set. S502: Based on the joint feature cycle number set, obtain the ventilation rhythm segment corresponding to each number, extract the inspiratory flow rate change trajectory, rhythm stage information and synchronization status identifier in the segment, determine whether the current rhythm cycle number meets the starting conditions for dynamic adjustment of oxygen supply according to the position of the current rhythm cycle number in the cycle sequence, add oxygen supply action type, rhythm stage mark and synchronization status data to the cycle segment that meets the conditions, and generate oxygen supply action parameter set. S503: Call the ventilation rhythm segment corresponding to each rhythm cycle in the oxygen supply action parameter set, insert the oxygen supply action parameters into the corresponding cycle segment structure in sequence, construct the edge control content item containing the oxygen supply action, integrate all cycle control units, arrange them in the rhythm sequence order to form a complete configuration dataset, and generate the edge communication cycle control configuration set. 2.The edge communication control method of a medical oxygen generation system according to claim 1, wherein: The ventilation rhythm segmentation includes the inspiratory phase coverage area, inspiratory flow rate time trajectory, respiratory action start and end markers, and rhythm cycle number. The expansion action time offset list includes the initial response timestamp, rhythm correspondence, offset time interval, and cycle offset sequence. The asynchronous cycle identifier set includes the response asynchronous cycle number, offset amplitude change trend, cycle continuity label, and asynchronous state classification result. The flow rate mutation cycle sequence includes the flow rate mutation cycle number, trend interruption type marker, rising segment change pattern, and mutation fluctuation amplitude. 3.The edge communication control method of a medical oxygen generation system according to claim 1, wherein: The edge communication cycle control configuration set includes oxygen supply action type, rhythm stage identifier, synchronization status parameters, control instruction content items, and communication cycle number.

Images