Method for monitoring abnormality of face velocity of fume hood

Abstract

The present application relates to the field of data processing, and particularly relates to an abnormal monitoring method for face wind speed of a fume hood, which comprises the following steps: defining a face wind area of an operator, and calculating an average wind condition vector of the wind area; then calculating a wind direction danger degree and a wind speed danger degree, and calculating a comprehensive danger coefficient of the wind area according to the wind direction danger degree and the wind speed danger degree; when the comprehensive danger coefficient is greater than or equal to a preset abnormal threshold, determining that the current face wind speed of the fume hood is abnormal, inputting the comprehensive danger coefficient into an improved PID, outputting an increment of a valve opening degree, adding the increment of the valve opening degree to a current valve opening degree value to obtain a new valve opening degree, and adjusting an exhaust capacity of the fume hood according to the new valve opening degree, so as to change the face wind speed of the fume hood. The present application calculates the comprehensive danger coefficient by taking the face wind area as a monitoring object, can more truly reflect the exposure of the operator in the actual operation process, and makes the monitoring result more practically meaningful.

Description

Technical Field

[0001] This invention relates to the field of data processing. More specifically, this invention relates to a method for anomaly monitoring of face velocity in a fume hood. Background Technology

[0002] Fume hoods play a vital role in chemical, biological, and other laboratory environments. They not only provide a relatively enclosed and safe environment for experimental operations but also effectively protect the health of personnel. Through a well-designed and controlled airflow system, fume hoods can remove harmful gases, particulate matter, volatile organic compounds, and other potentially toxic substances generated during experiments, preventing their spread within the laboratory space and avoiding harm to personnel or adverse effects on experimental equipment and reagents. This ensures the smooth conduct of experiments and the overall safety of the laboratory.

[0003] Currently, existing technologies, such as the patent application document with publication number CN119245176A, disclose a method and system for controlling the transmission and exhaust of a laboratory fume hood. In this method, the face velocity of the fume hood during laboratory operation is obtained, the ventilation area generated by the pull displacement sensor is calculated, the air volume is calculated based on the face velocity and the ventilation area, and the air volume is automatically adjusted by PID through the exhaust valve with a fast actuator on the fume hood.

[0004] However, when adjusting the opening of the exhaust valve, the proportional gain of a conventional PID controller is usually fixed. In actual operation, when the air velocity on the fume hood is close to the target value, the fixed proportional gain can easily cause the control system to overreact to small fluctuations, frequently issuing small adjustment commands, thereby causing unnecessary oscillations in the valve and airflow, making it impossible to accurately adjust the airflow and affecting experimental safety.

[0005] Therefore, in order to avoid the above situation, it is necessary to monitor the changes in face wind speed in real time and make dynamic adjustments to ensure that the face wind speed remains stable near the set value. Summary of the Invention

[0006] To address the limitations of using PID algorithms to adjust the opening of exhaust valves, which leads to inaccurate airflow control and affects experimental safety, this invention provides the following technical solution.

[0007] A method for monitoring abnormal face velocity in a fume hood, comprising: Construct a full-area three-dimensional wind field model of the laboratory; The operator's nose position is captured, and a facial wind zone is defined. The average wind condition vector of the wind zone is calculated using a three-dimensional wind field model. The wind direction hazard level and wind speed hazard level of the wind zone are calculated using the average wind condition vector of the wind zone. The comprehensive hazard coefficient of the wind zone is calculated based on the wind direction hazard level and wind speed hazard level. When the overall risk factor is greater than or equal to the preset abnormal threshold, the current face velocity of the fume hood is determined to be abnormal. Then, the proportional gain and integral gain are adjusted using the overall risk factor to obtain an improved PID. The overall risk factor is input into the improved PID, and the valve opening increment is output. The valve opening increment is added to the current valve opening value to obtain a new valve opening. The exhaust volume of the fume hood is adjusted according to the new valve opening, thereby changing the face velocity of the fume hood.

[0008] Preferably, the construction includes: Install wind speed sensors inside and outside the fume hood according to the preset layout requirements, and start all sensors to collect data. The collected data includes wind vectors for wind direction and wind speed. The entire interior space of the laboratory is taken as the wind field space and divided into regular grids. The intersection of the grids is the point to be reconstructed. For each point to be reconstructed in the wind field space, the Euclidean distance from it to each sensor is calculated, and the reciprocal of the Euclidean distance is taken as the weight coefficient of the corresponding sensor. Based on the wind vector and weighting coefficient of each sensor, the wind vector of each point to be reconstructed is calculated, and then a three-dimensional wind field model of the entire laboratory is reconstructed.

[0009] Preferably, the definition of the facial wind zone includes: The operator's facial images are continuously acquired, and the position of the operator's nose is determined by nose feature point detection. The nose is then mapped onto a three-dimensional wind field model, and the area with a preset radius centered on the nose position is taken as the wind zone of the operator's face.

[0010] Preferably, before calculating the wind direction hazard level and wind speed hazard level of the wind area using the average wind condition vector of the wind area, the method further includes: Define the horizontal inward normal vector of the fume hood. Perform a dot product operation between the average wind condition vector of the operator's face and the horizontal inward normal vector of the fume hood. If the result of the dot product operation is greater than or equal to 0, the current wind direction of the fume hood is determined to be safe; otherwise, if it is less than 0, the current wind direction of the fume hood is determined to be dangerous, and the degree of danger of the wind direction needs to be calculated.

[0011] Preferably, the degree of wind hazard is the negative cosine of the angle between the inward horizontal normal vector of the fume hood and the average wind condition vector of the face wind zone.

[0012] Preferably, the calculation process for the wind speed hazard level includes: The ideal face velocity of the preset fume hood; The ratio of the absolute value of the difference between the average wind speed vector of the operator's face and the ideal value of the face wind speed to the ideal value of the face wind speed is used as the wind speed hazard level of the face wind zone.

[0013] Preferably, when the cosine of the angle between the horizontal inward normal vector of the fume hood and the average wind condition vector of the face wind zone is greater than or equal to 0, the wind speed hazard level is used as the comprehensive hazard coefficient; when the cosine of the angle between the horizontal inward normal vector of the fume hood and the average wind condition vector of the face wind zone is less than 0, the product of the wind speed hazard level and the wind direction hazard level is used as the comprehensive hazard coefficient.

[0014] Preferably, adjusting the proportional gain using the comprehensive risk factor includes: Determine the initial proportional gain and the threshold for minor anomalies. Use the ratio of the comprehensive risk coefficient to the threshold for minor anomalies as the input to the hyperbolic tangent function for calculation. Use the product of the calculation result and the initial proportional gain as the adjusted proportional gain.

[0015] Preferably, adjusting the integral gain using the comprehensive risk factor includes: When the overall risk factor is greater than the minor anomaly threshold and the current valve opening is within the preset range, the pre-set integral gain is used; otherwise, the integral gain is 0.

[0016] The beneficial effects of this invention are: This invention first scientifically arranges wind speed sensors inside and outside the fume hood to comprehensively collect wind field data. After grid division and weight calculation, a three-dimensional wind field model of the entire laboratory is constructed to understand the normal distribution of wind speed on the fume hood surface and to build a basic framework for anomaly monitoring. Next, considering that one of the main functions of a fume hood is to protect operators from harmful gases and dust generated during the experiment, the operator's face, especially the mouth and nose, is the area directly exposed to the fume hood and most likely to inhale harmful substances. Therefore, advanced image recognition technology is used to accurately locate the operator's nose, determine the facial wind zone and map it onto a three-dimensional model, clarify the core area for monitoring abnormal face wind speed, and use the facial wind zone as the monitoring object to conduct in-depth analysis from the two key dimensions of wind direction and wind speed to calculate the comprehensive risk factor. This can more realistically reflect the operator's exposure during the actual operation process and make the monitoring results more meaningful. Finally, based on the comprehensive risk factor, an improved PID algorithm is used to dynamically adjust parameters and precisely control the valve opening to change the exhaust volume, forming a closed-loop feedback control mechanism based on anomaly monitoring to ensure that the fume hood surface velocity is always maintained within a safe and stable range. Attached Figure Description

[0017] Figure 1This is a flowchart of steps S1-S3 in a method for abnormal monitoring of face velocity in a fume hood according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0019] Reference Figure 1 A method for abnormal monitoring of face velocity in a fume hood includes steps S1-S3, as detailed below: S1: Construct a three-dimensional wind field model covering the entire laboratory area.

[0020] Given that traditional fume hoods only monitor the wind speed at a single point at the hood opening, which cannot fully reflect the wind field distribution in the operator's breathing zone, this operation aims to construct a three-dimensional dynamic wind field model of the laboratory space to provide a comprehensive data foundation for subsequent facial wind zone risk assessment.

[0021] In one embodiment, sensors are deployed at a height of 1.5 meters ± 0.1 meters above the ground outside the fume hood. This height corresponds to the human breathing zone and can accurately reflect the wind field conditions in the operator's environment. On this height plane, several equal-area areas are divided according to the actual size and layout of the laboratory. When dividing the areas, factors such as the location of the fume hood, the placement of equipment in the laboratory, and the range of personnel activity must be fully considered to ensure comprehensive coverage of the operator's breathing zone and avoid monitoring blind spots. An ultrasonic anemometer is installed at the center of each divided area. During installation, it is essential to ensure that the sensors are securely installed and that their horizontal and vertical alignment meets requirements to guarantee the accuracy of the collected data. Simultaneously, the three-dimensional coordinates of each sensor are recorded. Here, the z-coordinate of the sensor is exemplarily set at 1.5 meters ± 0.1 meters.

[0022] Inside the fume hood, an array of evenly distributed anemometers is installed according to the hood's dimensions and internal structure. A grid pattern or other suitable layout can be used to ensure comprehensive monitoring of airflow changes within the fume hood. Anemometers are installed inside the fume hood according to the determined array layout. Similarly, the installation quality of the sensors must be ensured, and the three-dimensional coordinates of each sensor are recorded. The z-coordinate of the sensor inside the fume hood is determined based on the sensor's actual height within the fume hood.

[0023] After all sensors are installed, activate them to begin collecting wind speed data. Ensure the sensors are functioning correctly and capable of collecting wind speed information accurately and in real time. The data collected by the sensors includes a wind vector corresponding to that location, which includes wind direction and wind speed.

[0024] The entire interior space of the laboratory is taken as the wind field space (i.e. the area to be reconstructed). This area is divided into regular grids at certain intervals, and the intersection of the grids is the point to be reconstructed.

[0025] Furthermore, for each point to be reconstructed in the wind field space, the Euclidean distance from the point to each sensor is calculated based on its coordinates and the three-dimensional coordinates of each sensor. The reciprocal of the calculated Euclidean distance is used as the weighting coefficient for each sensor, indicating the degree of reference for each sensor when the wind field is reconstructed at the point to be reconstructed. The closer a sensor is to a certain location in space, the higher the reference value of its measurement data for reconstructing the wind field state at that location.

[0026] Furthermore, based on the wind condition vector and weighting coefficient of each sensor, the wind condition vector of each point to be reconstructed is calculated, satisfying the following relationship: In the formula, The wind vector of the point to be reconstructed. For the first Wind vector data collected by each sensor For the first The weighting coefficients of each sensor This represents the total number of sensors.

[0027] Repeat the above calculation process for all points to be reconstructed in the wind field space to obtain the wind condition vector of each point, thereby reconstructing the three-dimensional wind field model of the entire laboratory area.

[0028] By comprehensively calculating the wind condition vector of the point to be reconstructed by considering the weighting coefficients of each sensor, the actual distribution of the wind field is reflected more comprehensively and accurately than using data from one or a few sensors alone. Sensors closer to the point to be reconstructed have a higher weight, and their data has a greater impact on the reconstruction results. Weighted fusion of wind condition vectors from multiple sensors fully utilizes the data information from each sensor, avoiding inaccurate reconstruction results caused by the failure or anomaly of a single sensor. Even if some sensor data has errors or is missing, data from other sensors can be supplemented and corrected through weighted fusion. Calculating the wind condition vector for each point to be reconstructed in the wind field space allows for the construction of a three-dimensional wind field model covering the entire laboratory area. This complete wind field model can intuitively display the spatial distribution characteristics and variation patterns of the wind field, providing a detailed and accurate data foundation for subsequent research and analysis.

[0029] S2: Capture the operator's nose position, define the facial wind zone, and calculate the average wind condition vector of the wind zone using a three-dimensional wind field model within the facial wind zone; use the average wind condition vector of the wind zone to calculate the wind direction hazard level and wind speed hazard level of the wind zone, and calculate the comprehensive hazard coefficient of the wind zone based on the wind direction hazard level and wind speed hazard level.

[0030] Furthermore, considering that traditional fume hood safety assessment methods only use the scalar value of the air velocity at the hood opening as an indicator, ignoring the actual exposure risk to the operator's face, the three-dimensional wind field model of S1 described above quantifies the wind velocity deviation and wind direction hazard in the operator's facial area, which can more comprehensively and accurately reflect the actual operating status and ventilation effect of the fume hood.

[0031] In one embodiment, facial images of the operator are continuously acquired by a camera (ensuring clear images, uniform lighting, and minimizing background interference and obstructions). The position of the operator's nose is determined using nasal feature point detection methods such as traditional image processing-based nose tip localization methods, facial key point detection algorithms, or deep learning-based semantic segmentation methods. This nose is then mapped onto a three-dimensional wind field model. A hemispherical region with a radius of 15 centimeters centered on this position is used as the wind zone around the operator's face, thereby accurately focusing on the wind field environment in which the operator's face is located.

[0032] Further, the wind vector at each point within the facial wind zone is integrally divided by the volume of the facial wind zone to obtain the average wind vector of the facial wind zone, which reflects the average wind speed and direction at the operator's face.

[0033] In one embodiment, the inward horizontal normal vector of the fume hood is determined. This direction is typically determined based on the fume hood's design and related structural parameters. For example, if the front of the fume hood is the operating window, the inward horizontal normal vector is the horizontal vector pointing from the operating window towards the center area inside the fume hood.

[0034] Further, the average wind vector of the operator's facial airflow zone is multiplied by the determined inward normal vector of the fume hood. If the dot product is greater than or equal to 0, it indicates that the wind direction towards the inside of the fume hood is reasonable, and gas can be drawn into the fume hood from the operator's face. The current wind direction can be considered relatively safe. If the dot product is less than 0, it indicates that the wind direction towards the operator's face is not towards the inside of the fume hood but towards the operator's face; the wind direction is unreasonable, and there is a risk of gas being inhaled by the operator.

[0035] Based on the preliminary assessment above, when the result of the dot product operation is less than 0, the wind direction hazard level of the wind zone is further calculated. That is, the following relationship is satisfied: In the formula, The wind direction and hazard level of the facial wind zone. This represents the inward normal vector of the fume hood. This represents the average wind condition vector for the face wind zone. It is the cosine of the angle between the horizontal inward normal vector of the fume hood and the average wind condition vector of the face wind zone.

[0036] Since the result of the dot product operation at this time is Since it is less than 0, the result is obtained as follows: It is also less than 0, thus making The range of its value is (0, 1). The closer it is to 1, the more dangerous the current wind direction is, which means the more serious the deviation between the wind condition vector of the face wind zone and the normal vector direction of the fume hood inward.

[0037] In one embodiment, the ideal face velocity of the fume hood is first determined in the fume hood's technical documentation or operating specifications. The ratio of the absolute value of the difference between the average wind speed vector of the operator's face and the ideal face velocity to the ideal face velocity is used as the wind speed hazard level for that face area. The closer this hazard level is to 1, the more severe the deviation between the actual face velocity and the ideal face velocity, the more dangerous the current wind speed, and the ventilation effect may not meet expectations, posing a risk of harmful gas leakage or diffusion.

[0038] The calculated wind speed hazard level is combined with the calculated wind direction hazard level to comprehensively assess the safety of the ventilation system. Both reflect the ventilation effect of the fume hood in actual operation and the degree of potential risk faced by the operator.

[0039] In one embodiment, when the cosine of the angle between the horizontal inward normal vector of the fume hood and the average wind condition vector of the face wind zone is greater than or equal to 0, the wind speed hazard level is directly used as the comprehensive hazard coefficient, and only the influence of wind speed is considered; when the cosine of the angle between the horizontal inward normal vector of the fume hood and the average wind condition vector of the face wind zone is less than 0, the product of the wind speed hazard level and the wind direction hazard level is used as the comprehensive hazard coefficient.

[0040] S3: When the overall risk factor is greater than or equal to the preset abnormal threshold, the current fume hood face velocity is determined to be abnormal. Then, the proportional gain and integral gain are adjusted using the overall risk factor to obtain an improved PID. The overall risk factor is input into the improved PID, and the valve opening increment is output. The valve opening increment is added to the current valve opening value to obtain a new valve opening. The exhaust volume of the fume hood is adjusted according to the new valve opening, thereby changing the fume hood face velocity.

[0041] In one embodiment, when the comprehensive risk coefficient finally obtained by S2 is less than the preset abnormal threshold (in this embodiment of the invention, the abnormal threshold is set to 0.05, because in actual application scenarios, when the comprehensive risk coefficient reaches a certain small value, the system state can meet the basic requirements for stable operation), the fume hood system is considered to be in a stable state. At this time, the fume hood face velocity is relatively stable. Under these circumstances, the fume hood face velocity can be maintained at a relatively stable level without large fluctuations, thereby ensuring that the operator will not face risks caused by abnormal face velocity during operation.

[0042] However, when the overall risk factor obtained by S2 is greater than or equal to the preset abnormal threshold, the fume hood system is considered to be in an unstable state and the valve opening needs to be adjusted.

[0043] In the application scenario of this invention, the valve opening is physically limited. Traditional PID (Proportional-Integral-Derivative) controllers output the final control quantity (valve opening). Once the calculated opening exceeds the physical range, limiting is required, which may lead to system response lag. Incremental PID controllers, on the other hand, output the increment of the control quantity (the change in valve opening), adjusting only a small increment based on the current state each time. Even if the accumulated calculated increments might exceed the range, limiting each increment prevents valve opening oversaturation, thus better adapting to the valve opening limitation requirements.

[0044] In one embodiment, the comprehensive anomaly index finally calculated in S2 above is used as the core input of the incremental PID controller. This is because the comprehensive hazard coefficient can comprehensively reflect the wind speed and wind direction safety in the operator's facial area. The goal is to stabilize the face wind speed of the fume hood near the ideal value while ensuring wind direction safety. The comprehensive hazard coefficient, as an input, is directly related to this goal. By dynamically adjusting the valve opening, the wind speed and wind direction safety in the facial area are optimized, causing the comprehensive hazard coefficient to approach 0, thus achieving the ideal state.

[0045] In one embodiment, the proportional gain is dynamically adjusted based on the overall risk factor, specifically as follows: In the formula, This is the adjusted proportional gain. The initial proportional gain setting (set during initial debugging of the controller system). The comprehensive risk factor calculated using S2 above, This is a preset threshold for minor anomalies (e.g., 0.1). It is the hyperbolic tangent function.

[0046] When the system faces significant risks (the overall risk factor is much greater than the threshold of minor anomalies, such as when the overall risk factor is 0.5), The closer the proportional gain is to 1, the closer the adjusted proportional gain is to the initial proportional gain. The proportional link maintains strong control, driving the valve to adjust significantly and quickly reduce risk. Conversely, when the system is close to stable (the overall risk coefficient is less than the threshold of minor anomalies, such as when the overall risk coefficient is 0.05), the proportional gain will be less than 1. along with The decrease is approximately linear. This results in... When the gain is very small, the reduced proportional gain makes the control system less sensitive to minor anomalies, avoiding frequent fine-tuning and unnecessary oscillations of the valve when it is close to the ideal state, thus improving stability.

[0047] Considering that when the valve opening reaches its physical limits (0% fully closed or 100% fully open), even if the integral term detects a continuous deviation ( The integral term (>0) accumulates and increases continuously, making it impossible to further change the valve opening to achieve more airflow adjustment. When changes in external conditions lead to increased risk and require reverse valve adjustment, the control system must first "digest" this ineffective accumulated integral value before issuing an effective reverse control command because the integral term has already accumulated a large positive value. This results in system response delays, recovery lags, and an inability to respond to new risk situations in a timely manner.

[0048] In one embodiment, to prevent the integral term from accumulating ineffectively when the valve is at its limit opening and there is a deviation, eliminate the control lag caused by this, ensure that the valve can be effectively adjusted immediately once it leaves the saturation zone, and speed up the system response, the conditional integral term is activated based on the current valve opening and the comprehensive risk factor.

[0049] When the overall risk factor is greater than the minor anomaly threshold and the current opening degree is in the range of 20%-80%, the integral term is activated, that is, the integral gain value set through the initial debugging of the controller system is used to accumulate deviations normally to eliminate possible steady-state errors; when the condition that the overall risk factor is greater than the minor anomaly threshold and the current opening degree is in the range of 20%-80% is not met, the integral term is 0.

[0050] That is, the integral term will only be recorded to eliminate static error if both the existence of a significant anomaly (the overall risk factor is greater than the threshold of minor anomalies) and the valve is in the unsaturated zone (the current opening degree is in the range of 20%-80%). Otherwise, the integral term will be disabled.

[0051] Furthermore, based on the aforementioned adjustments to the proportional gain and integral gain, an improved incremental PID is obtained. The incremental valve opening is output using the improved incremental PID. The current valve opening is then added to the output valve opening increment to obtain the new valve opening. Based on the new valve opening, the fume hood control system adjusts the valve opening. As the valve opening changes, the exhaust volume of the fume hood also changes accordingly, thereby changing the face velocity. This stabilizes the face velocity of the fume hood near the ideal value, enabling the system to achieve stable operation and providing a safe and reliable working environment for the operator.

[0052] Specifically, the valve opening of a fume hood directly determines its ventilation cross-sectional area. When the valve opening increases, the ventilation cross-sectional area increases, allowing more air to enter the room. Conversely, when the valve opening decreases, the ventilation cross-sectional area decreases, and the amount of air entering the room decreases accordingly. The face velocity of a fume hood refers to the airflow velocity at the fume hood opening. When the exhaust volume increases, the face velocity will increase given a fixed fume hood opening area; conversely, when the exhaust volume decreases, the face velocity will decrease.

[0053] In summary, by monitoring the wind speed and direction in the operator's breathing zone, a comprehensive risk factor is calculated. Based on this factor, it is determined whether the face velocity of the fume hood is abnormal, and the valve opening of the fume hood is dynamically adjusted to control the exhaust volume, thereby ensuring that the wind speed and direction in the operator's face area are always within a safe range.

[0054] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for monitoring abnormal face velocity in a fume hood, characterized in that, include: Construct a full-area three-dimensional wind field model of the laboratory; The system captures the operator's nose position, defines the facial wind zone, and calculates the average wind condition vector within this zone using a 3D wind field model. It then uses this average wind condition vector to calculate the wind direction and speed hazard levels for that zone, including: The formula for calculating the degree of hazard of the wind direction is: In the formula, The wind direction and hazard level of the facial wind zone. This represents the inward normal vector of the fume hood. This represents the average wind condition vector of the facial wind zone; It is the cosine of the angle between the horizontal inward normal vector of the fume hood and the average wind condition vector of the face wind zone; The calculation process for the wind speed hazard level includes: setting the ideal value of the face wind speed of the fume hood; and taking the absolute value of the difference between the wind speed value of the average wind condition vector of the operator's face wind zone and the ideal value of the face wind speed as the wind speed hazard level of the face wind zone. The comprehensive hazard coefficient for the wind zone is calculated based on the hazard levels of wind direction and wind speed. This includes: when the cosine of the angle between the horizontal inward normal vector of the fume hood and the average wind condition vector of the face wind zone is greater than or equal to 0, the wind speed hazard level is used as the comprehensive hazard coefficient; when the cosine of the angle between the horizontal inward normal vector of the fume hood and the average wind condition vector of the face wind zone is less than 0, the product of the wind speed hazard level and the wind direction hazard level is used as the comprehensive hazard coefficient. When the overall risk factor is greater than or equal to the preset abnormal threshold, the current face velocity of the fume hood is determined to be abnormal. Then, the proportional gain and integral gain are adjusted using the overall risk factor to obtain an improved PID. The overall risk factor is input into the improved PID, and the valve opening increment is output. The valve opening increment is added to the current valve opening value to obtain a new valve opening. The exhaust volume of the fume hood is adjusted according to the new valve opening, thereby changing the face velocity of the fume hood.

2. The method for abnormal monitoring of face velocity in a fume hood according to claim 1, characterized in that, The construction of the full-domain three-dimensional wind field model of the laboratory includes: Install wind speed sensors inside and outside the fume hood according to the preset layout requirements, and start all sensors to collect data. The collected data includes wind vectors for wind direction and wind speed. The entire interior space of the laboratory is taken as the wind field space and divided into regular grids. The intersection of the grids is the point to be reconstructed. For each point to be reconstructed in the wind field space, the Euclidean distance from it to each sensor is calculated, and the reciprocal of the Euclidean distance is taken as the weight coefficient of the corresponding sensor. Based on the wind vector and weighting coefficient of each sensor, the wind vector of each point to be reconstructed is calculated, and then a three-dimensional wind field model of the entire laboratory is reconstructed.

3. The method for abnormal monitoring of face velocity in a fume hood according to claim 1, characterized in that, The defined facial wind zone includes: The operator's facial images are continuously acquired, and the position of the operator's nose is determined by nose feature point detection. The nose is then mapped onto a three-dimensional wind field model, and the area with a preset radius centered on the nose position is taken as the wind zone of the operator's face.

4. The method for abnormal monitoring of face velocity in a fume hood according to claim 1, characterized in that, Before calculating the wind direction and wind speed hazard levels of the wind area using the average wind condition vector, the following steps are also included: Define the horizontal inward normal vector of the fume hood. Perform a dot product operation between the average wind condition vector of the operator's face and the horizontal inward normal vector of the fume hood. If the result of the dot product operation is greater than or equal to 0, the current wind direction of the fume hood is determined to be safe; otherwise, if it is less than 0, the current wind direction of the fume hood is determined to be dangerous, and the degree of danger of the wind direction needs to be calculated.

5. The method for abnormal monitoring of face velocity in a fume hood according to claim 1, characterized in that, Adjusting the proportional gain using the overall risk factor includes: Determine the initial proportional gain and the threshold for minor anomalies. Use the ratio of the comprehensive risk coefficient to the threshold for minor anomalies as the input to the hyperbolic tangent function for calculation. Use the product of the calculation result and the initial proportional gain as the adjusted proportional gain.

6. The method for abnormal monitoring of face velocity in a fume hood according to claim 5, characterized in that, Adjusting the integral gain using the overall risk factor includes: When the overall risk factor is greater than the minor anomaly threshold and the current valve opening is within the preset range, the pre-set integral gain is used; otherwise, the integral gain is 0.

Images