Negative pressure ward multi-parameter dynamic sensing control monitoring method based on Internet of Things

Abstract

The invention relates to the technical field of sensors, in particular to a negative pressure ward multi-parameter dynamic sensing control monitoring method based on the Internet of Things, which comprises the following steps of: acquiring differential pressure personnel access control particle concentration through LoRa to construct a differential pressure time sequence, estimating slope turning by least square to extract disturbance parameters, and calculating a differential pressure time sequence; collecting the air speed of an air outlet in combination with disturbance parameters, calculating an air supply and exhaust included angle to identify an abnormal airflow region, extracting particle concentration data of the corresponding region, identifying a coverage interruption point by using a CUSUM algorithm, and obtaining an access control state and people flow density calculation difference to judge a multi-parameter abnormal monitoring point; and an abnormal attribution membership degree is calculated by associating the number intersection of the airflow anomaly and the multi-parameter anomaly point, an anomaly type is recognized again in combination with an attribution threshold value, and a dynamic monitoring result is output. According to the method, multi-source dynamic cooperative monitoring is carried out by extracting pressure difference slope turning identification trend disturbance, calculating air supply and exhaust angles to position an airflow abnormal area, recognizing coverage weak points in combination with particle concentration fluctuation, and fusing access control and people flow density difference values to mark risk points.

Description

Technical Field

[0001] This invention relates to the field of sensor technology, and in particular to a multi-parameter dynamic sensor monitoring method for negative pressure wards based on the Internet of Things. Background Technology

[0002] The field of sensor technology encompasses the detection and conversion of environmental, physical, physiological, and chemical parameters. Its core principle is to utilize physical or chemical effects to convert various non-electrical signals into electrical signals for the identification, acquisition, and transmission of target parameters. Based on the multiple objects being detected, it can be categorized into temperature, humidity, pressure, gas composition, and biological characteristics, with applications covering industrial automation, healthcare, environmental monitoring, and intelligent transportation. In healthcare settings, sensor technology continuously monitors air quality, pathogen transmission conditions, and physiological data, providing fundamental information support for safety control and management, and is a crucial component of the Internet of Things (IoT) sensing layer.

[0003] Among them, the IoT-based multi-parameter dynamic sensing and control monitoring method for negative pressure wards refers to the use of multiple environmental and physiological sensors under the IoT architecture to monitor and collect data on multiple key parameters inside the negative pressure ward in real time. The technical aspects include using a barometric pressure sensor to monitor changes in indoor and outdoor pressure difference to determine the negative pressure maintenance status, using an air quality sensor to monitor particulate matter and harmful gas concentrations to help determine the degree of air purification, using a temperature and humidity sensor to collect thermal and humidity environment data to support microenvironment regulation, and uploading the above data to the backend server through a low-power wireless network protocol via an embedded communication terminal, and completing dynamic data recording and storage by combining timestamps.

[0004] Current technologies rely on single-sensor air pressure and air quality detection, using static thresholds to determine pressure differential maintenance and particulate matter concentration. This provides only a rough assessment of the overall environment, lacking in-depth analysis of dynamic trends and fluctuations. It often fails to distinguish between transient disturbances and persistent anomalies caused by short-term personnel entry / exit or access control actions. Monitoring of supply and exhaust air nodes often neglects wind speed and direction, making it impossible to accurately locate abnormal airflow areas and affecting the accuracy of subsequent disinfection coverage and environmental control. The evaluation of disinfection particle coverage lacks time-series-based error analysis, relying heavily on fixed sampling points, easily overlooking temporary coverage blind spots. Correlating access control status with pedestrian density through isolated data processing fails to comprehensively determine the interaction between the environment and personnel activities, making it difficult to optimize flow control and disinfection strategies in a targeted manner. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a multi-parameter dynamic infection control monitoring method for negative pressure wards based on the Internet of Things.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a multi-parameter dynamic infection control and monitoring method for negative pressure wards based on the Internet of Things, comprising the following steps: S1: Obtain the pressure difference of the negative pressure ward channel uploaded through the LoRa communication protocol and construct the pressure difference time series. Fit the pressure difference time series using the least squares method and calculate the pressure difference fluctuation frequency to generate a pressure difference fluctuation state sequence. S2: Based on the pressure difference fluctuation state sequence, collect the exhaust outlet wind speed and calculate the angle between the wind speed of each region's air supply node and the direction of the adjacent node. Combine the pressure difference fluctuation frequency to screen the airflow areas with abnormal disturbances and generate an abnormal airflow guidance area set. S3: By collecting the air particle concentration change value in the corresponding area during the disinfection process through the airflow guidance anomaly zone set, the effectiveness of the air particle wave range is identified, the cumulative offset of the concentration drop segment is identified by the CUSUM algorithm, it is determined whether the concentration fluctuation threshold is exceeded, and a disinfection coverage interruption point set is generated. S4: Based on the disinfection coverage interruption point set, obtain the corresponding access control status record and people flow density through the Internet of Things edge gateway, calculate the difference between the access control opening and closing time and the people flow density in the same area, and determine whether it exceeds the disturbance threshold value, and generate a multi-parameter abnormal monitoring point set. S5: Based on the set of multi-parameter abnormal monitoring points and the set of airflow-directed abnormal areas, calculate the membership degree of the abnormality at the intersection of the numbers, re-identify the current abnormality type in combination with the abnormality membership threshold, and output the set of multi-parameter dynamic disinfection and infection control monitoring results for negative pressure wards.

[0007] As a further embodiment of the present invention, the differential pressure fluctuation state sequence specifically includes regression coefficients, residual offset values, and differential pressure fluctuation frequencies; the airflow guidance anomaly zone set includes region number, air supply direction angle value, and anomaly disturbance identifier; the disinfection coverage interruption point set specifically refers to the interruption point time sequence, interruption point region identifier, and cumulative offset; the multi-parameter anomaly monitoring point set specifically includes monitoring point number, population density difference, and access control duration difference; and the negative pressure ward multi-parameter dynamic disinfection and infection control monitoring result set includes region anomaly type label, membership degree, and monitoring result timestamp.

[0008] As a further aspect of the present invention, the specific steps of S1 include: S101: Obtain the pressure difference of the uploaded negative pressure ward channel, receive the raw pressure difference data packet transmitted by the LoRa node and parse out the timestamp and the corresponding pressure difference value, convert it to the standard time format according to the timestamp, arrange multiple data points in ascending order of time and fill and connect them to construct pressure difference time series data; S102: Based on the pressure difference time series data, the pressure difference change trend is fitted by the least squares method, the fitting error of each segment is calculated one by one according to the time period, and the set of numerical offset difference between the fitted value and the original value of the time period is obtained to form the pressure difference fitting offset value group. S103: Based on the pressure difference fitting offset value group, mark the turning points of the offset values, measure the time interval between consecutive turning points, and back-calculate the fluctuation frequency within the time period according to the interval. Then, classify and summarize the corresponding frequencies in chronological order to generate a pressure difference fluctuation state sequence.

[0009] As a further aspect of the present invention, the specific steps of S2 include: S201: Based on the pressure difference fluctuation state sequence, collect the wind speed at the exhaust outlet of the negative pressure ward area, pair the wind speed measurement value with the corresponding measuring point coordinates, and number and summarize them in chronological order to establish a set of wind speed values ​​at the air supply point. S202: Call the wind speed value set of the air supply point, calculate the angle between the wind speed vectors according to the coordinates between the air supply node and the adjacent node in the area, and record the corresponding node number and the size of the angle, obtain the relative flow direction information between multiple air supply points, and establish a group of air supply direction angle values. S203: Based on the air supply direction angle value group, combined with the pressure difference fluctuation state sequence, filter the area points where the angle change exceeds the angle difference threshold within the corresponding time period, classify and summarize all areas where directional disturbances and frequency anomalies overlap, and generate an airflow guidance anomaly area set. The angle difference threshold is set by selecting wind speed direction data from multiple regions under the condition that the pressure difference fluctuation frequency is less than 0.1 Hz as a benchmark, and statistically analyzing the fluctuation amplitude of the angle between adjacent nodes.

[0010] As a further aspect of the present invention, the specific steps of S3 include: S301: Through the airflow-guided abnormal zone set, collect the air particulate monitoring point data set in the corresponding area during the disinfection process, extract the PM2.5 and PM10 particulate concentration values ​​corresponding to each monitoring point, summarize them in time order and establish a multi-monitoring point particle concentration change record to obtain a particulate concentration change value group. S302: Based on the particle concentration change value group, compare the particle concentration fluctuation amplitude between multiple monitoring points in multiple regions, identify abnormal area nodes in the fluctuation range within the same time period, and count the overlap of the wave time between the node and the adjacent area and filter out the coverage failure area to obtain the effective range of particle wave. S303: Call the particle wave and effectiveness interval, use the CUSUM algorithm to calculate the cumulative offset value of the continuous decreasing segment in the particle concentration time series, combine the concentration fluctuation threshold to determine whether the offset value exceeds the limit, mark the time period that meets the condition as the state change point, and generate a disinfection coverage interruption point set. The concentration fluctuation threshold is set by statistically analyzing the decreasing trend of air particulate concentration under normal continuous coverage conditions during disinfection operations.

[0011] As a further aspect of the present invention, the specific steps of S4 include: S401: Based on the disinfection coverage interruption point set, retrieve the access control status data and people flow sensor data recorded by the IoT edge gateway, extract the access control opening and closing indicators and people counts in the corresponding time period of the area where multiple interruption points are located, and organize them into interval statistical results to generate access control status and people flow density data groups. S402: Call the access control status and people density data group, respectively calculate the total duration of continuous access control opening during the interruption period of multiple areas and the people density during the same time period, convert the people density value according to the area standard, construct a set of parallel indicators of people density and access control opening and closing time, and establish a set of disturbance difference values ​​in the same area. S403: Based on the set of disturbance difference values ​​in the same area, combined with the access control duration and pedestrian density limit in the set disturbance threshold, filter the area number that meets any condition, and mark and summarize all the interruption point locations that meet the conditions to generate a multi-parameter anomaly monitoring point set.

[0012] As a further aspect of the present invention, the disturbance threshold value is set by selecting normal time period data from several regions and calculating the distribution range of access control opening duration and the number of people.

[0013] As a further aspect of the present invention, the specific steps of S5 include: S501: Based on the set of multi-parameter abnormal monitoring points and the set of airflow guidance abnormal areas, match and retrieve the two types of area numbers, identify the set of points with overlapping spatial locations, extract the number index and associated parameter group corresponding to the intersection area, and obtain the index value group of abnormal intersection area. S502: Call the index value group of the abnormal intersection region, calculate the superposition intensity between the multi-parameter abnormal indicators and the airflow indicators in the multi-intersection region and normalize it, calculate the membership degree value of the multi-region, and identify the region category that exceeds the limit by combining the abnormal membership threshold to obtain the abnormal membership degree interval. S503: Based on the abnormal membership interval and combined with the classification characteristics under multiple parameter indicators, re-determine the current abnormal classification label and update the regional status identifier. Output the status results in chronological order and correspondingly with the regional number to generate a multi-parameter dynamic disinfection and infection control monitoring result set for negative pressure wards.

[0014] As a further aspect of the present invention, the anomaly attribution threshold is statistically set by constructing the membership degree distribution of multi-parameter anomaly samples. The intersection region where the attribution category is determined under multiple anomaly types is selected as a reference sample. The normalized score of the corresponding multi-dimensional index under the cross state is calculated and a membership degree distribution curve is formed. The critical value corresponding to the intersection point of accurate classification rate and misclassification rate is extracted and set.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, trend disturbance parameters are extracted by constructing a channel pressure difference time series and the number of slope inflection points to achieve a keen perception of pressure changes inside negative pressure wards, assisting in the identification of stage fluctuations and reducing false detection rates. Based on the trend disturbance parameters, exhaust vent wind speed is simultaneously collected, and the angle between the wind direction of multiple air supply nodes and adjacent nodes is calculated, which can accurately locate abnormal airflow guidance areas and narrow the monitoring range. Moving average error analysis is performed on the concentration of disinfectant particles in abnormal areas, and weak points are covered by fluctuation threshold marking, realizing dynamic identification and real-time feedback of disinfection effects. By combining the difference between the access control opening and closing time and the flow density corresponding to the abnormal points, a multi-parameter abnormal monitoring point set is generated to realize the coupled analysis of environmental factors and personnel activities, providing precise support for disinfection strategies and flow control optimization. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the main steps of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise specifically defined.

[0019] Please see Figure 1 This invention provides a technical solution: a multi-parameter dynamic infection control monitoring method for negative pressure wards based on the Internet of Things, comprising the following steps: S1: Obtain the pressure difference of the negative pressure ward channel uploaded through the LoRa communication protocol and construct the pressure difference time series. Fit the pressure difference time series using the least squares method and calculate the pressure difference fluctuation frequency to generate a pressure difference fluctuation state sequence. S2: Based on the differential pressure fluctuation state sequence, the exhaust outlet wind speed is collected and the angle between the wind speed of each air supply node and the direction of the adjacent node is calculated. Combined with the differential pressure fluctuation frequency, the airflow areas with abnormal disturbances are screened to generate an abnormal airflow guidance area set. S3: By collecting the air particle concentration change value in the corresponding area during the disinfection process through the airflow-guided abnormal area set, the effectiveness of the air particle wave range is identified, the cumulative offset of the concentration drop segment is identified by the CUSUM algorithm, it is determined whether the concentration fluctuation threshold is exceeded, and a disinfection coverage interruption point set is generated. S4: Based on the disinfection coverage interruption point set, obtain the corresponding access control status records and people flow density through the IoT edge gateway, calculate the difference between the access control opening and closing time and the people flow density in the same area, and determine whether it exceeds the disturbance threshold value to generate a multi-parameter abnormal monitoring point set; S5: Based on the set of multi-parameter abnormal monitoring points and the set of airflow-oriented abnormal areas, calculate the membership degree of the abnormality at the intersection of the numbers, and re-identify the current abnormality type by combining the abnormality membership threshold, and output the set of multi-parameter dynamic disinfection and infection control monitoring results of the negative pressure ward.

[0020] The differential pressure fluctuation state sequence specifically includes regression coefficients, residual offset values, and differential pressure fluctuation frequency. The airflow guidance anomaly zone set includes area number, air supply direction angle value, and abnormal disturbance identifier. The disinfection coverage interruption point set specifically refers to the interruption point time series, interruption point area identifier, and cumulative offset. The multi-parameter abnormal monitoring point set specifically includes monitoring point number, population density difference, and access control duration difference. The multi-parameter dynamic disinfection and infection control monitoring result set for negative pressure wards includes area abnormality type label, membership degree, and monitoring result timestamp.

[0021] Please see Figure 1 The specific steps for obtaining S1 are as follows: S101: Obtain the pressure difference of the uploaded negative pressure ward channel, receive the raw pressure difference data packet transmitted by the LoRa node and parse out the timestamp and the corresponding pressure difference value, convert it to the standard time format according to the timestamp, arrange multiple data points in ascending order of time and fill and connect them to construct pressure difference time series data; The hexadecimal data packet 0xA3D5C720250718132515.2 transmitted from the LoRa node is received. The last 12 characters, 202507181325, are extracted as a timestamp, which is parsed as 13:25 on July 18, 2025. The time zone offset is added by 8 hours to convert it to UTC standard time 05:25. The pressure difference value corresponding to the timestamp 202507181330 of the adjacent node is obtained as 18.6 Pa. A 5-minute gap in the time interval is detected. Three data points are inserted between 13:25 and 13:30. The pressure difference value at 13:27 is calculated using linear interpolation. Pa, 13:28 Pa generates a continuous time series (see Table 1).

[0022] Table 1. Complete table of pressure difference time series Timestamp Original value (Pa) Complete value (Pa) 2025-07-18T13：25 15.2 - 2025-07-18T13：27 - 16.3 2025-07-18T13：28 - 17.5 2025-07-18T13：30 18.6 - As shown in Table 1, the data sequence was constructed by timestamp parsing and interpolation calculation, forming a complete dataset with minute intervals from 13:25 to 13:30.

[0023] S102: Based on the time series data of differential pressure, the trend of differential pressure change is fitted by the least squares method. The fitting error of each segment is calculated one by one according to the time period. The set of numerical offset difference between the fitted value and the original value of the time period is obtained to form the differential pressure fitting offset value group. The 24-hour data was divided into 48 30-minute time intervals. Representing the The time-period error value is calculated by summing the absolute deviations of all measuring points in the numerator and introducing a balance parameter in the denominator. To prevent error distortion when there are too few measuring points, Representing the Time period The raw differential pressure value at the measuring point is converted using the 16-bit AD sampling value transmitted by the LoRa node. The conversion formula is as follows: ( (Voltage signal of 0-3.3V) The value represents the least squares fitted value, derived from the trend equation. calculate( For time variables, (fit coefficients) This represents the number of valid measurement points, calculated after removing outlier data with signal strength below -110dBm. Represents the equilibrium parameter, according to calculate( (Indicates rounding down), when hour Raw value acquisition: AD sampling values ​​of measurement point i=3 during the time period j=5 (06:00-06:30). Calculated The equation was obtained by fitting the time series of 15 measuring points within the time period. ( (in minutes, at t=28 hours): Absolute deviation calculation The total deviation of 15 measuring points Pa, substitute into the formula to calculate the error value: Threshold determination, setting an error threshold (Based on the differential pressure control requirements in GB50019-2015 "Code for Design of Heating, Ventilation and Air Conditioning of Industrial Buildings"), because The fitting results for the time period were deemed valid, and the time period was identified as a high-error period.

[0024] S103: Based on the pressure difference fitting offset value group, mark the inflection point of the offset value, measure the time interval between consecutive inflection points, and back-calculate the fluctuation frequency within the time period according to the interval. Then, classify and summarize the corresponding frequencies in chronological order to generate a pressure difference fluctuation state sequence. In the offset value sequence [1.2, 1.4, 1.6, 1.3, 0.9]Pa, the threshold for the turning point is set as the change in adjacent points. Pa, the change in pressure at point 3 (1.6 Pa) from point 2 (1.4 Pa) was detected to be +0.2 Pa, and the change at point 4 (1.3 Pa) from point 3 was detected to be -0.3 Pa. Point 4 was marked as the turning point. The time interval between adjacent turning points was calculated: when the turning points occurred at 09:15 and 11:30, the time difference was 135 minutes. The fluctuation frequency was then calculated. The frequency is categorized into the 0.4-0.5Hz band, and the cumulative number of occurrences in this band is incremented by 1. The differential pressure fluctuation state sequence is interconnected with the airflow monitoring system to provide dynamic pressure change input for airflow guidance analysis.

[0025] Please see Figure 1 The specific steps to obtain S2 are as follows: S201: Based on the pressure difference fluctuation state sequence, collect the air velocity at the exhaust outlet of the negative pressure ward area, pair the measured air velocity values ​​with the corresponding measuring point coordinates, and number and summarize them in chronological order to establish a set of air supply point air velocity values. The raw data packet ID (5A3D, 2025-07-18, T13:25:00Z), X: 23, Y: 45, V: 2.3 was obtained from the FS5-FlowSens ultrasonic anemometer installed at the exhaust vent. The X coordinate (23 m / s), Y coordinate (45 m / s), and wind speed value (2.3 m / s) were extracted. The timestamp 2025-07-18, T13:25:00Z was converted to a sequence. Column number 20250718132505, when the data ID: N7, 2025-07-18T13:25:10Z, X: 25, Y: 47, V: 2.5 sent by the adjacent node N7 at 13:25:10 is detected, the continuity of time is verified. If the number interval exceeds 5 seconds, a virtual data point is inserted, including inserting data at 13:25:07 between 13:25:05 and 13:25:10. The wind speed value is taken as the average value. m / s, coordinates taken from the midpoint , This forms a time-series dataset (see Table 2).

[0026] Table 2. Partial dataset of air supply point wind speeds Timestamp X-coordinate (m) Y coordinate (m) Wind speed (m / s) 2025-07-18T13：25：00 23 45 2.3 2025-07-18T13：25：07 24 46 2.4 2025-07-18T13：25：10 25 47 2.5 As shown in Table 2, a complete wind speed dataset with 12 sampling points per minute was constructed through time alignment and interpolation.

[0027] S202: Call the wind speed value set of the air supply point, calculate the angle between the wind speed vectors based on the coordinates between the air supply node and the adjacent node in the area, record the corresponding node number and the size of the angle, obtain the relative flow direction information between multiple air supply points, and establish a group of air supply direction angle values. Select node A1 (coordinates) ) and adjacent node A2 ( ), calculate vector Node A3 ( )vector Calculate the dot product , module length , The included angle is obtained. When the angle between nodes A1-A2-A3 changes from 115.2° to 62.3° (a change of 52.9°) between 13:25 and 13:30, the node group number [A1, A2, A3] and the angle change value are recorded.

[0028] S203: Based on the air supply direction angle value group, combined with the pressure difference fluctuation state sequence, filter the area points where the angle change exceeds the angle difference threshold within the corresponding time period, classify and summarize all areas with overlapping directional disturbances and frequency anomalies, and generate an airflow guidance anomaly area set. The angle difference threshold is set by selecting wind speed direction data from multiple regions under the condition that the pressure difference fluctuation frequency is less than 0.1 Hz as a benchmark and statistically analyzing the fluctuation amplitude of the angle between adjacent nodes. During the time period of 0.4-0.5Hz pressure difference fluctuation (13:25-13:35), the corresponding air supply angle data is extracted. Angle difference thresholds are set according to Clause 5.3.2 of GB / T13554-2020 "Code for Construction and Acceptance of Cleanrooms". If a change in the angle of node group [B4, B5, B6] is detected at 13:28, the system first retrieves the list of area numbers from the airflow guidance anomaly area database. The database table stores fields including area number, detection timestamp, and angle change value. For example, when querying timestamp 13:28:00, area numbers B4, B5, and B6 are retrieved. The data is presented in numerical format. The system stores data in groups [B4, B5, B6]. It sends HTTP requests to the particle monitoring system via a predefined application programming interface (API). The request method is POST, and the request header includes Content-Type: application / json and Authorization: Bearertoken123. The request body is in JSON format, including an array of region IDs and a timestamp of 13:28:00. Upon receiving the request, the particle monitoring system parses the JSON data, extracts the region IDs, and queries each ID from the region configuration database. The corresponding physical coordinates are configured in the database table, which includes the region number, X coordinate, Y coordinate, and elevation. For example, region B4 corresponds to X=10.5 meters, Y=20.3 meters, and elevation=1.2 meters. The coordinate data is obtained through total station measurement, using the reference point (0, 0, 0), with a measurement error of ±0.05 meters. Region B5 corresponds to X=15.2 meters and Y=25.1 meters, and region B6 corresponds to X=20.0 meters and Y=30.5 meters. The system also collects real-time particle concentration data from a particle sensor network. Sensor data is published at a frequency of 1 Hz via the MQTT protocol. For sensor / particle, the payload format is {sensor-id: P1, x: 10.0, y: 20.0, concentration: 55, timestamp: 13:28:00}. Sensor P1 coordinates (10.0, 20.0), concentration 55 particles / m³; sensor P2 coordinates (15.0, 25.0), concentration 60 particles / m³; sensor P3 coordinates (20.0, 30.0), concentration 58 particles / m³. The system calculates the Euclidean distance between each region and the nearest sensor. For region B4, the distance calculation is... The distance is approximately 0.58 meters. Compared with the preset sensor association threshold of 1.0 meter, 0.58 < 1.0. Therefore, sensor P1 is associated, and its concentration value of 55 particles / m³ is used for calculation based on the distance to area B5. ≈0.22 meters, 0.22 < 1.0, associated sensor P2, concentration value 60 particles / m³, distance calculation for area B6. =0.50 m, 0.50 < 1.0, associated sensor P3, concentration value 58 particles / m³, the system compares the concentration value with the particle concentration threshold of 50 particles / m³, the threshold is set by statistically analyzing historical normal data, the data is the particle concentration value array of the same period in the past 30 days [38, 42, 45, 40, 39, 41, 43, 44, 42, 40, 45, 46, 41, 39, 38, 40, 42, 44, 43, 41, 39, 40, 42, 45, 46, 44, 43, 41, 40, 39], (38+42+45+40+39+41+43+44+42+40+45+46+41+39+38+40+42 +44+43+41+39+40+42+45+46+44+43+41+40+39) / 30=41.4 units / m³. To calculate the variance, first calculate the sum of squared differences: =[(38-41.4)²+(42-41.4)²+(45-41.4)²+(40-41.4)²+(39-41.4)²+(41-41.4)²+(43-41.4)²+(44-41.4)²+(42-41.4)²+(40-41.4)²+(45-41.4)²+(46-41.4)²+(41-41.4)²+(39-41.4)²+(38-41.4)²+(40-41.4)²+(40-41.4)²+(40-41.4)²+(42-41.4)²+(42-41.4)²+(40-41.4)²+(45-41.4)²+(46-41.4)²+(41-41.4)²+(39-41.4)²+(38-41.4)²+(40-41.4)²+(40-41.4)²+(41-41.4)²+(39-41.4)²+(40-41.4)²+(41-41.4)²+(40-41.4)²+(41-41.4)²+(41-41.4)²+(39 ... 1.4)²+(42-41.4)²+(44-41.4)²+(43-41.4)²+(41-41.4)²+(39-41.4)²+(40-41.4)²+(42-41.4)²+(45-41.4)²+(46-41.4)²+(44-41.4)²+(43-41.4)²+(41-41.4)²+(40-41.4)²+(39-41.4)²]≈150.8, variance=150.8 / 30≈5.027, standard deviation=2.24 / m³, threshold set as mean plus 2 times standard deviation=41.4+2×2.24=45.88, rounded to 50 In the example, the concentration of B4 is 55 > 50, the concentration of B5 is 60 > 50, and the concentration of B6 is 58 > 50. Since the concentrations in all areas exceed the threshold, the system generates a list of abnormal region coordinates in the format {region: B4, x: 10.5, y: 20.3}, {region: B5, x: 15.2, y: 25.1}, and {region: B6, x: 20.0, y: 30.5}. This list is sent to the disinfection effect analysis module via a WebSocket connection. The analysis module parses the coordinate data and drives the disinfection equipment to move. For example, the ultraviolet disinfection equipment moves from its initial position (0, 0) to coordinates (10.5, 20.3) in B4, and the moving distance is calculated. The distance is approximately 22.85 meters, the device speed is set to 0.5 meters per second, and the movement time is 22.85 / 0.5 = 45.7 seconds. The system records the movement log, including the starting coordinates, target coordinates, and movement time. The system repeats the above data calls, API requests, coordinate queries, distance calculations, concentration comparisons, list generation, and data transmission actions. The system completes the correlation and positioning of the abnormal airflow area and particle monitoring data. The area coordinates, sensor correlation data, and distance calculations are used for system communication and positioning processing. At the same time, the pressure difference fluctuation frequency of 0.47 Hz exceeds the system threshold of 0.45 Hz, and the area Z-09 is marked as an abnormal area. When 3 overlapping abnormalities are detected in the same hour, Z-09 is added to the abnormal area set, and alarm code AL-20250718132809 is generated.

[0029] Please see Figure 1 The specific steps to obtain S3 are as follows: S301: By using the airflow-guided abnormal zone set, collect data from the air particulate monitoring points set in the corresponding area during the disinfection process, extract the PM2.5 and PM10 particulate concentration values ​​corresponding to each monitoring point, summarize them in chronological order and establish a record of particle concentration changes at multiple monitoring points to obtain a group of particulate concentration change values. The laser particle counter data packet 2025-07-18, T14:00:00, PM2.5=12, PM10=23 was obtained from monitoring point P7 in anomaly zone Z-09. The timestamp 14:00:00 was extracted. When the next data packet 2025-07-18, T14:05:00, PM2.5=15, PM10=28 was detected, a 5-minute time interval gap was found. The data point 14:02:30 was inserted between 14:00 and 14:05, and the PM2.5 value was taken as the average. μg / m³, PM10 μg / m³, for missing data due to equipment failure (including data loss at 14:07:30), linear prediction is used based on the first two valid data points, assuming... PM2.5 = 15 If PM2.5 = 18, then the interpolation value at 14:07:30 is... μg / m³, forming a complete time series record (see Table 3).

[0030] Table 3. Particle Concentration Dataset (Excerpt) Timestamp PM2.5 (μg / m³) PM10 (μg / m³) 2025-07-18T14：00：00 12 23 2025-07-18T14：02：30 13.5 25.5 2025-07-18T14：05：00 15 28 2025-07-18T14：07：30 16.5 30.2 As shown in Table 3, a complete dataset with 2 sampling points per minute is constructed through time alignment and interpolation.

[0031] S302: Based on the particle concentration change value group, compare the particle concentration fluctuation amplitude between multiple monitoring points in multiple regions, identify abnormal area nodes in the fluctuation range within the same time period, and count the overlap of the wave time between the node and the adjacent area and filter out the coverage failure area to obtain the effective range of particle wave. Data from region A (monitoring points A1-A3) and region B (B1-B3) during the period from 14:00 to 14:15 were selected to calculate the multi-point PM2.5 fluctuation range: standard deviation of region A. μg / m³, region B μg / m³, set an abnormal threshold μg / m³ (according to the hospital area limit in GB3095-2012 "Ambient Air Quality Standard"), area B is judged to be abnormal. The time difference between the fluctuation start time of area B (14:05) and the fluctuation start time of the adjacent area C (14:07) is 2 minutes, which is less than the preset overlap threshold of 5 minutes, so it is judged as a successfully covered area. When the fluctuation start time of area D (14:10) is 6 minutes different from the surrounding areas, it is marked as a failed coverage area.

[0032] S303: Call the particulate wave effectiveness interval, use the CUSUM algorithm to calculate the cumulative offset value of the continuous decreasing segment in the particulate concentration time series, combine the concentration fluctuation threshold to determine whether the offset value exceeds the limit, mark the time period that meets the condition as the state change point, and generate a set of disinfection coverage interruption points. The concentration fluctuation threshold is set by statistically analyzing the decreasing trend of air particulate concentration under normal continuous coverage conditions during disinfection operations. Represents the cumulative offset metric. Representing the The minute-by-minute PM2.5 concentration value was obtained by averaging the samples taken every 30 seconds using a TSI8530 sensor. The average value represents the baseline. Data from 120 samples taken during normal operation of the toxic gas fan was excluded from the calculation, as was data from the equipment calibration period. Represents the concentration difference between adjacent time points, calculated as follows: Theoretically, the effective disinfection value should be negative. Represents the maximum single-step change, taken from 30 days of data. The maximum value is 8 μg / m³. The standard deviation represents the normal range of fluctuation. The measurement error compensation term is set to 0.5 based on the sensor accuracy of ±0.5μg / m³ to prevent the denominator from being too small. The PM2.5 data sequence [35, 32, 29, 27, 25, 23] μg / m³ during the normal detoxification operation period (14:00-14:30) is used as the mean. Calculate the standard deviation Maximum absolute difference Measurement error (Based on sensor accuracy ±0.5μg / m³), adjacent differences , Calculate the CUSUM value when u=3: ,set up (According to ISO 14644-1 Cleanroom Standard Class 7 requirements), when the cumulative value exceeds the threshold An alarm is triggered when the value of disinfection is 0.646 < 2.0, indicating that there was no interruption in disinfection during that time period. At that time, alarm AL-20250718143002 was triggered.

[0033] Please see Figure 1 The specific steps to obtain S4 are as follows: S401: Based on the disinfection coverage interruption point set, retrieve the access control status data and people flow sensor data recorded by the IoT edge gateway, extract the access control opening and closing indicators and people counts in the corresponding time period of the area where multiple interruption points are located, and organize them into interval statistical results to generate access control status and people flow density data groups. Access control records from the edge gateway are read at 2025-07-18, T14:30:00, Gate7, Open and 2025-07-18, T14:45:00, Gate7, Close. The opening duration is calculated to be 15 minutes. Data packets from the AX-09 pedestrian flow sensor in this area are simultaneously acquired at 14:30:00, Count=5; 14:35:00, Count=8; 14:40:00, Count=12. When a missing data point is detected at 14:33:00, the average of the values ​​before and after is taken. People are interpolated at a rate of 0.5 people per minute to generate a complete sequence [5, 5.5, 6, 6.5, 7, 7.5, 8, 9, 10, 11, 12]. The area is 82 square meters (based on architectural floor plan CAD-2024-07). The calculated pedestrian density at 14:35 is as follows. For each person / m², for access control anomaly records (including status word verification failure at 14:32:00), the previous 5 valid records are traced back to repair the status. When there are 3 consecutive read failures, the data is marked as invalid (see Table 4).

[0034] Table 4 Access Control and Pedestrian Flow Data Groups (Excerpt) Timestamp Access control status Number of people Area (㎡) 2025-07-18，T14：30：00 Open 5 82 2025-07-18，T14：35：00 Open 8 82 2025-07-18，T14：40：00 closure 12 82 As shown in Table 4, a complete time series record is constructed through data repair and interpolation.

[0035] S402: Call the access control status and people density data group, respectively calculate the total duration of continuous access control opening during the interruption period of multiple areas and the people density during the same time period, convert the people density value according to the area standard, construct a set of parallel indicators of people density and access control opening and closing time, and establish a set of disturbance difference values ​​in the same area. Data was selected for area BX-12 between 14:00 and 15:00, and the access control was continuously open for 47 minutes (14:10-14:57). The peak number of people was 21. The area is 105㎡. The peak density was calculated. People / m², with baseline values ​​set as follows: normal access control duration ≤ 30 minutes (based on WS233-2017 "Technical Specifications for Hospital Isolation"), and a people density threshold of 0.15 people / m² (based on GB51039-2014 "Code for Design of General Hospital Buildings"). Disturbance difference is calculated as: access control duration difference. Minutes, density difference For people / m², when the data for area CX-08 is access control difference +5 minutes and density difference -0.02, construct parallel index sets [+17, +0.05] and [+5, -0.02]. Take the absolute value of the negative difference to generate the disturbance set [17, 0.05, 5, 0.02].

[0036] S403: Based on the set of disturbance difference values ​​in the same area, combined with the access control duration and crowd density limit in the set disturbance threshold, filter the area number that meets any condition, and mark and summarize all the interruption point locations that meet the conditions to generate a multi-parameter anomaly monitoring point set. Two thresholds were set: access control duration difference ≥ 10 minutes or density difference ≥ 0.03 people / ㎡. The data of the detection area DX-15 was access control duration difference +12 minutes and density difference +0.04, which met the dual conditions to trigger an alarm. The data of the area EX-20 was access control duration difference +8 minutes and density difference +0.05, with only density exceeding the standard. When the access control duration difference of the area FX-03 was +15 minutes (exceeding the threshold by 50%) and density difference +0.01, it was marked as a single condition anomaly. The results were summarized to generate alarm codes [AL-20250718144512, AL-20250718145005], which were pushed to the BIM operation and maintenance platform to display the red warning area. The results show that there is a risk of coupling between access control failure and personnel gathering.

[0037] Please see Figure 1 The specific steps to obtain S5 are as follows: S501: Based on the multi-parameter anomaly monitoring point set and the airflow guidance anomaly area set, the two types of area numbers are matched and retrieved to identify the spatially overlapping point set, extract the number index and associated parameter group corresponding to the intersection area, and obtain the index value group of the anomaly intersection area. Extract regional coordinate data from the multi-parameter anomaly alarm set [AL-20250718143002, AL-20250718144512] and the airflow anomaly set [AL-20250718132809, AL-20250718143015]. Analyze the planar coordinates (X: 23.45m, Y: 45.78m, based on the BIM model origin coordinate system) of region Z-09 corresponding to AL-20250718143002, and the coordinates (X: 23.46m, Y: 45.77m) of region Z-09 corresponding to AL-20250718143015. Calculate the Euclidean distance between the two points. Areas with a distance less than 1.0 meter (according to Clause 6.2.3 of GB / T35648-2017 "Classification and Coding Standard for Building Information Modeling") are marked as completely overlapping areas. For partially offset areas such as AL-20250718144512 (X: 25.12m, Y: 47.33m) and AL-20250718132809 (X: 24.95m, Y: 46.88m), the distance is calculated. The value of the meter is still less than the threshold. The relevant parameter group is extracted as follows: the change in airflow angle of Z-09 is 42.9°, the access control timeout is 17 minutes, and the PM2.5 concentration fluctuation is +4.8μg / m³ (see Table 5).

[0038] Table 5 Key Parameters of Abnormal Intersection Regions Area code Changes in airflow angle (°) Access control timeout (minutes) PM2.5 fluctuations (μg / m³) Z-09 42.9 17 +4.8 CX-08 38.2 5 +3.1 As shown in Table 5, the multidimensional parameters of the spatially overlapping regions have been structurally integrated.

[0039] S502: Call the index value group of abnormal intersection region, calculate the superposition intensity between multi-parameter abnormal indicators and airflow indicators in the multi-intersection region and normalize it, calculate the membership degree value of multiple regions, and identify the region category exceeding the limit by combining the abnormal membership threshold to obtain the abnormal membership degree interval. For area Z-09, the following weightings were assigned: abnormal airflow direction weight 0.6 (according to Clause 4.3.5 of JGJ / T454-2019 "Technical Standard for Operation and Maintenance of Ventilation and Air Conditioning Systems in Medical Buildings"), abnormal access control weight 0.3, and abnormal particulate concentration weight 0.1. The superposition intensity was then calculated. The system's maximum intensity value of 58.7 over the past 30 days (recorded on July 15, 2025) was obtained and normalized. Set the attribution threshold to 0.5 (taking the 90th percentile of the data), and calculate the intensity value of region CX-08 as 0.412. When, the normalization result Values ​​below the threshold were excluded. Z-09, exceeding the threshold of 0.533, was marked as a Level 1 anomaly. An intensity value of 0.61 was detected in the FX-03 area (airflow 53.1° × 0.6 + access control 22 minutes × 0.3 + PM2.5 fluctuation 5.3 × 0.1 = 31.86 + 6.6 + 0.53 = 38.99, normalized). The anomaly levels are divided into 0.1 intervals: 0.5-0.6 is Level II (yellow warning), and >0.6 is Level I (red warning). FX-03 is now updated to Level I anomaly.

[0040] S503: Based on the membership interval of the abnormality and combined with the classification characteristics under multiple parameter indicators, re-determine the classification label of the current abnormality and update the regional status identifier. Output the status results in chronological order and corresponding to the regional number to generate a set of multi-parameter dynamic disinfection and infection control monitoring results for negative pressure wards. The characteristic parameters analyzed for Level I anomaly FX-03 were: airflow angle change of 51.2° (exceeding the threshold of 45°), access control timeout of 22 minutes (exceeding the threshold of 15 minutes), and PM2.5 fluctuation of +5.3 μg / m³ (exceeding the threshold of 5.0). Based on the multi-parameter coupling rule in Article 8.2 of WS / T367-2012 "Technical Specifications for Disinfection in Medical Institutions," it was reclassified as a composite airflow-personnel control failure. Simultaneously, Level II anomaly Z-09 (airflow 42.9° / access control timeout of 17 minutes / PM2.5 +4.8) was marked as an airflow-dominant anomaly. The time series monitoring results were output as: 2025-0 On July 18, 2025, at 14:45:00, a red alert was triggered in area FX-03. On July 18, 2025, at 14:30:00, a yellow alert was triggered in area Z-09. The dynamic indicators of the building operation and maintenance platform were updated simultaneously: FX-03 displayed a red pulse warning box, and Z-09 displayed an orange static border. Records were stored in ISO8601 time format, including four core fields: area number, anomaly level, classification label, and parameter set. The monitoring result set was subjected to pressure monitoring, airflow analysis, and particle monitoring and interconnected to generate a closed-loop sensing and control data chain, forming a complete monitoring result dataset.

[0041] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A negative pressure ward multi-parameter dynamic sensing and monitoring method based on the Internet of Things, characterized in that, The method comprises the following steps: S1: obtaining the pressure difference of the uploaded negative pressure ward channel through the LoRa communication protocol and constructing a pressure difference time sequence, fitting the pressure difference time sequence through the least square method and calculating the pressure difference fluctuation frequency to generate a pressure difference fluctuation state sequence; S2: based on the pressure difference fluctuation state sequence, collecting the air outlet wind speed and calculating the angle between the wind speed of each regional air supply node and the adjacent node direction, combining the pressure difference fluctuation frequency to screen the airflow region of abnormal disturbance, and generating an airflow guide abnormal region set; S3: through the airflow guide abnormal region set, collecting the air particle concentration change value of the corresponding region in the disinfection process, identifying the effectiveness of the air particle spread range, identifying the cumulative offset of the concentration decline section through the CUSUM algorithm, judging whether it exceeds the concentration fluctuation threshold, and generating a disinfection coverage breakpoint set; S4: based on the disinfection coverage breakpoint set, obtaining the corresponding access control state record and people flow density through the Internet of Things edge gateway, calculating the difference between the access control opening and closing time and the people flow density in the same region respectively, and judging whether it exceeds the disturbance threshold, and generating a multi-parameter abnormal monitoring point set.

2. The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 1, characterized in that, The pressure difference fluctuation state sequence is specifically a regression coefficient, a residual offset value and a pressure difference fluctuation frequency, the airflow guide abnormal region set comprises a region number, an air supply direction angle value and an abnormal disturbance identifier, the disinfection coverage breakpoint set specifically refers to a breakpoint time sequence, a breakpoint region identifier and a cumulative offset, and the multi-parameter abnormal monitoring point set specifically comprises a monitoring point number, a people flow density difference value and an access control time difference value. 3.The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 1, characterized in that, The specific steps of S1 comprise: S101: obtaining the uploaded pressure difference of the negative pressure ward channel, receiving the original pressure difference data packet transmitted by the LoRa node and parsing the timestamp and the corresponding pressure difference value, converting the timestamp to a standard time format according to the timestamp, arranging the multiple data points in ascending order of time, filling and connecting, and constructing a pressure difference time sequence data; S102: based on the pressure difference time sequence data, fitting the pressure difference change trend through the least square method, calculating the fitting error of each section one by one according to the time period, obtaining the numerical offset difference value set between the time period fitting value and the original value, and forming a pressure difference fitting offset value group; S103: according to the pressure difference fitting offset value group, marking the turning point of the offset value, measuring the time interval between the continuous turning points, inversely calculating the fluctuation frequency in the time period according to the interval, classifying and summarizing the corresponding frequency in time sequence, and generating a pressure difference fluctuation state sequence.

4. The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 1, characterized in that, The specific steps of S2 comprise: S201: based on the pressure difference fluctuation state sequence, collecting the air outlet wind speed of the negative pressure ward region, pairing the wind speed measurement value with the corresponding measurement point coordinates, numbering and summarizing in time sequence, and establishing a wind speed value set of air supply points; S202: calling the wind speed value set of air supply points, calculating the angle value between the wind speed vectors according to the coordinates between the air supply nodes and the adjacent nodes in the region, recording the corresponding node number and angle size, obtaining the relative flow direction information between multiple air supply points, and establishing an air supply direction angle value group; S203: According to the angle difference threshold value, the angle difference threshold value is set by selecting the wind speed direction data of multiple regions under the condition that the pressure difference fluctuation frequency is lower than 0.1 Hz as a reference, and the fluctuation amplitude of the included angle between adjacent nodes is counted. The specific steps of S3 include:

5. The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 1, characterized in that, S301: Through the airflow guiding abnormal area set, the air particle monitoring point data in the corresponding area during the disinfection process is collected, the PM2.5 and PM10 particle concentration values corresponding to each monitoring point are extracted, the particle concentration change records of multiple monitoring points are summarized and established in time sequence, and the particle concentration change value group is obtained; S302: Based on the particle concentration change value group, the particle concentration fluctuation amplitudes between multiple monitoring points in multiple regions are compared, the abnormal region nodes in the fluctuation interval within the same time period are identified, and the coincidence of the wave time of the nodes and the adjacent regions is counted and the coverage failure area is screened, and the particle wave effectiveness interval is obtained; S303: The particle wave effectiveness interval is called, the cumulative offset value of the continuous descending segment in the particle concentration time sequence is calculated by using the CUSUM algorithm, whether the offset value exceeds the limit is judged combined with the concentration fluctuation threshold value, the time period meeting the conditions is marked as a state mutation point, and the disinfection coverage breakpoint set is generated. The concentration fluctuation threshold value is set by counting the downward trend of the air particle concentration under the condition of normal continuous coverage of the disinfection operation. The specific steps of S4 include:

6. The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 1, characterized in that, S401: Based on the disinfection coverage breakpoint set, the access control state data and people flow sensor data recorded by the Internet of Things edge gateway are called, the access control opening and closing identifier and the number of people in the region where multiple breakpoints are located within the corresponding time period are extracted, and the interval statistical results are arranged to generate the access control state and people flow density data group; S402: The access control state and people flow density data group is called, the total time length of the access control continuously opened in the interruption period in multiple regions and the people flow density in the same time period are respectively counted, the people flow density value is converted according to the area standard, the parallel index set of people flow density and access control opening and closing time length is constructed, and the same region disturbance difference set is established; S403: According to the same region disturbance difference set, the access control time length and people flow density limit in the set disturbance threshold value are combined, the region numbers meeting any condition are screened, and all the breakpoint positions meeting the conditions are marked and summarized to generate the multi-parameter abnormal monitoring point set. The disturbance threshold value is set by selecting the normal period data of several regions to calculate the distribution range of the access control opening duration and the number of people.

7. The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 6, characterized in that, The method further includes: 8.The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 1, wherein, S5: Based on the multi-parameter abnormal monitoring point set and the airflow guiding abnormal area set, the abnormal attribution membership degree of the intersection region is calculated, the current abnormal type is re-identified combined with the abnormal attribution threshold value, and the negative pressure ward multi-parameter dynamic disinfection and infection control monitoring result set is output; The negative pressure ward multi-parameter dynamic disinfection and infection control monitoring result set includes region abnormal type label, membership degree, and monitoring result timestamp. ​ 9.The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 1, wherein, The specific steps of S5 include: S501: Based on the multi-parameter abnormal monitoring point set and the air flow guiding abnormal area set, the two types of region numbers are matched and searched, the point set with spatial position coincidence is identified, the number index and the associated parameter group corresponding to the intersection region are extracted, and the abnormal intersection region index value group is obtained; S502: Calling the abnormal intersection region index value group, the superposition strength between the multi-parameter abnormal index and the air flow index in the multi-intersection region is calculated and normalized, the belonging membership value of the multi-region is calculated, the region category exceeding the limit is identified combined with the abnormal belonging threshold, and the abnormal belonging membership interval is obtained; S503: According to the abnormal belonging membership interval, combined with the classification characteristics under the multi-parameter index, the classification label of the current abnormality is re-determined and the region state identifier is updated, the state result is correspondingly output according to the time sequence and the region number, and the multi-parameter dynamic disinfection and control monitoring result set of the negative pressure ward is generated.

10. The IoT-based multi-parameter dynamic monitoring method for a negative pressure ward according to claim 9, characterized in that, The abnormal belonging threshold is statistically set by constructing the belonging membership distribution of the multi-parameter abnormal sample, the intersection region belonging to the determined labeled belonging category under multiple abnormal types is selected as the reference sample, the normalized score of the corresponding multi-dimensional index in the cross state is calculated and the membership distribution curve is formed, and the critical value corresponding to the intersection point of the accurate classification rate and the misjudgment rate is extracted and set.

Images