Air purification control method and system based on neural network

Abstract

The invention relates to the technical field of air purification, in particular to an air purification control method and system based on a neural network, and the method comprises the following steps: S1, environment analysis: setting basic threshold values of various pollutants according to environment types and environment ranges; s2, character analysis: analyzing people flow and character behaviors in the environment, and setting a dynamic adjustment coefficient according to the average people flow and the current people flow; analyzing the type and concentration of pollutants possibly generated according to the character behaviors, and setting behavior influence parameters; s3, dynamic threshold calculation: dynamically adjusting the basic threshold according to the dynamic adjustment coefficient and the behavior influence parameter to obtain a dynamic threshold; and S4, air dynamic regulation and control, wherein the concentration of pollutants in the environment is monitored, and the concentration of the pollutants and a dynamic threshold value of the pollutants are compared. According to the method, the dynamic adjustment coefficient and the behavior influence parameter are generated by predicting the influence of the human flow and the human behavior on the air, the basic threshold value is dynamically adjusted, and the air purification efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of air purification technology, specifically to an air purification control method and system based on neural networks. Background Technology

[0002] With rapid industrialization and urbanization, and increased human activity, indoor air quality problems have become increasingly prominent. Long-term exposure to polluted environments can lead to respiratory diseases, cardiovascular problems, allergic reactions, and even increase the risk of cancer. World Health Organization (WHO) data shows that approximately 7 million people worldwide die prematurely each year from diseases caused by air pollution, with indoor air pollution accounting for more than 50% of these deaths. Therefore, air purification technology plays a crucial role in protecting human health and improving the quality of living and working environments.

[0003] Existing air purification devices, such as the Blueair Classic 480i smart air purifier, are equipped with PM2.5, VOCs, and temperature and humidity sensors to detect the concentration of various pollutants in the environment. By setting preset thresholds, when the concentration of pollutants reaches the threshold, the device will respond immediately (e.g., automatically increase the fan speed when PM2.5 > 35 μg / m³) to optimize the air quality.

[0004] However, existing air purification methods rely excessively on fixed thresholds, lack air quality prediction, cannot dynamically adapt to environmental changes, and have low purification efficiency. Therefore, it is necessary to propose a neural network-based air purification control method and system that can predict air quality changes, dynamically adjust thresholds, and improve purification efficiency. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides an air purification control method and system based on neural networks. By designing dynamic adjustment coefficients and behavioral influence parameters, the method predicts the impact of pedestrian traffic and behavior on air quality based on the amount of pedestrian traffic and behavior in the environment, generates dynamic adjustment coefficients and behavioral influence parameters, and dynamically adjusts the basic threshold to adapt to environmental changes and improve air purification efficiency.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: an air purification control method based on neural networks, comprising the following steps: S1, Environmental Analysis: Real-time acquisition of environmental images; analysis of environmental type and scope based on environmental images; setting basic thresholds for various pollutants based on environmental type and scope; and setting different types and levels of early warning prompts and control schemes.

[0007] S2, People Analysis: Based on real-time collected environmental images, analyze the average and current pedestrian flow and people's behavior in the environment, and set dynamic adjustment coefficients based on the average and current pedestrian flow; analyze the types and concentrations of pollutants that people may generate based on their behavior, and set behavior impact parameters.

[0008] S3, Dynamic Threshold Calculation: Based on the dynamic adjustment coefficient and the behavior influence parameter, the basic threshold is dynamically adjusted to obtain the dynamic threshold.

[0009] S4, Dynamic Air Control: Monitors the concentration of various pollutants in the environment, compares the concentration of each pollutant with its dynamic threshold; when the concentration of any pollutant exceeds its corresponding dynamic threshold, issues a warning of the corresponding type and level based on the type of pollutant and the amount exceeded, and initiates the corresponding control scheme.

[0010] Furthermore, the early warning system includes Level 1, Level 2, and Level 3 warnings, with the warning level being directly proportional to the pollutant concentration.

[0011] Furthermore, an air purification control system based on neural networks includes an image acquisition module, an environmental recognition module, a human recognition module, a dynamic analysis module, and a dynamic control module.

[0012] The image acquisition module includes an image recording device; the image acquisition module is used to acquire real-time images of the current environment using the image recording device, generate live images, and transmit the live images to the environment recognition module and the person recognition module in real time.

[0013] The environment recognition module receives real-time images acquired by the image acquisition module, extracts environmental features from the real-time images, constructs an environment recognition model based on a neural network algorithm, analyzes the environmental features using the environment recognition model, obtains the environment type and environmental range, and transmits the obtained environment type and environmental range to the dynamic analysis module.

[0014] The person recognition module receives live images captured by the image acquisition module, extracts person images from the live images, constructs a person recognition model based on computer vision and deep learning algorithms, analyzes person characteristics using the person recognition model, obtains average pedestrian flow, current pedestrian flow, and person behavior in the environment, predicts the type and concentration of pollutants generated by person behavior, and generates behavior impact parameters; and transmits the average pedestrian flow, current pedestrian flow, and behavior impact parameters to the dynamic analysis module.

[0015] The dynamic analysis module receives the environmental type and scope from the environmental identification module, analyzes typical pollutants within the environmental type, maps corresponding monitoring and control equipment, generates an equipment installation report, and transmits the report to the dynamic control module. It also sets basic thresholds for each pollutant based on the environmental scope. Furthermore, it receives current pedestrian traffic and behavioral impact parameters from the people and people identification module, sets dynamic adjustment coefficients based on average and current pedestrian traffic, calculates the current dynamic thresholds of the environment in real time based on the dynamic adjustment coefficients and behavioral impact parameters, and transmits these dynamic thresholds to the dynamic control module in real time.

[0016] The dynamic control module includes several monitoring and control devices. It receives equipment installation reports and dynamic thresholds from the dynamic analysis module and transmits them to the user terminal, allowing the user to install monitoring and control devices in the environment based on the monitoring equipment installation reports. It also uses the monitoring devices to monitor the concentration of various pollutants in the environment in real time, generate real-time concentrations, and compares the real-time concentrations with the dynamic thresholds. If the real-time concentration is greater than the dynamic threshold, the dynamic control module controls the operation of the corresponding control devices.

[0017] Furthermore, when the dynamic threshold ≤ real-time concentration < 110% of the dynamic threshold, the dynamic control module is used to issue a Level 1 warning; when the dynamic threshold ≤ real-time concentration < 130% of the dynamic threshold, the dynamic control module is used to issue a Level 2 warning; and when the dynamic threshold ≤ real-time concentration, the dynamic control module is used to issue a Level 3 warning.

[0018] Furthermore, the formula for calculating the dynamic threshold is as follows: △T=T0·KP (1).

[0019] Where △T is the dynamic threshold, T0 is the basic threshold, K is the dynamic adjustment coefficient, the value of K ranges from 0.1 to 2, and P is the behavior influence parameter.

[0020] Furthermore, the formula for calculating the dynamic adjustment coefficient is as follows: K=N p / N N (2).

[0021] Where, N p N represents the average foot traffic over the past 7 days. N N represents the current foot traffic. When the foot traffic is 0, N represents the current foot traffic. N Take 1.

[0022] Furthermore, environmental features include environmental outlines and environmental landmarks, while human features include human outlines, human postures, and objects carried.

[0023] Furthermore, the monitoring equipment includes several gas concentration detectors.

[0024] Furthermore, the control equipment includes several air filters, exhaust fans, air intake ducts, and air exhaust ducts.

[0025] Furthermore, the control methods of the control scheme include activating the air filter, adjusting the exhaust fan speed, and adjusting the number of opening and closing of the air intake and exhaust ducts.

[0026] The above approach has the following beneficial effects: 1. Existing air purification methods rely excessively on fixed thresholds, lack air quality prediction, cannot dynamically adapt to environmental changes, and have low purification efficiency. In contrast to existing air purification methods, this solution can monitor the flow of people and their behavior in the environment in real time. Based on the flow of people and their behavior, it can predict the impact of these factors on air quality, generate dynamic adjustment coefficients and behavior impact parameters, and dynamically adjust the basic threshold to adapt to environmental changes and improve air purification efficiency.

[0027] 2. Different types and ranges of environments have varying tolerance to different pollutants; existing air purification methods lack consideration for environmental type and range when designing thresholds; compared to existing air purification methods, this solution can collect real-time environmental images and analyze the environmental type and range based on these images, allowing users to set more reasonable basic thresholds according to the environmental type and range. This enables the purification method to better purify air for different environments, improving the solution's adaptability to the environment and its air purification effect.

[0028] 3. This solution, through the design of dynamic analysis and dynamic control modules, can effectively analyze changes in current human traffic and their impact on the environment. It also analyzes existing human behavior and its impact on the environment, adjusting dynamic thresholds in real time. The dynamic control module then compares the dynamic thresholds with the concentrations of various pollutants in the current environment. Based on the comparison results, it issues different levels of warnings and executes corresponding control measures, thereby achieving appropriate treatment for pollution of different severity levels. While ensuring the effectiveness of treatment, it avoids overtreatment, improving the efficiency and rationality of treatment.

[0029] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0030] Figure 1 This is a schematic diagram illustrating the steps of the air purification control method based on neural networks according to the present invention.

[0031] Figure 2 A schematic diagram of the structure of the air purification control system based on neural networks of this invention. Detailed Implementation

[0032] The following detailed description illustrates the specific implementation method: Example 1:

[0033] As attached Figure 1 As shown, an air purification control method based on neural networks includes the following steps: S1, Environmental Analysis: Real-time acquisition of environmental images; analysis of environmental type and scope based on environmental images; setting basic thresholds for various pollutants based on environmental type and scope; and setting different types and levels of early warning prompts and control schemes.

[0034] S2, People Analysis: Based on real-time collected environmental images, analyze the average and current pedestrian flow and people's behavior in the environment, and set dynamic adjustment coefficients based on the average and current pedestrian flow; analyze the types and concentrations of pollutants that people may generate based on their behavior, and set behavior impact parameters.

[0035] S3, Dynamic Threshold Calculation: Based on the dynamic adjustment coefficient and the behavior influence parameter, the basic threshold is dynamically adjusted to obtain the dynamic threshold.

[0036] S4, Dynamic Air Control: Monitors the concentration of various pollutants in the environment, compares the concentration of each pollutant with its dynamic threshold; when the concentration of any pollutant exceeds its corresponding dynamic threshold, issues a warning of the corresponding type and level (warnings include Level 1, Level 2 and Level 3, with the warning level being proportional to the pollutant concentration) based on the type of pollutant and the amount exceeding the threshold, and activates the corresponding control scheme.

[0037] The specific experiment is as follows: I. Experimental Preparation Experimental group: Air purification and early warning were performed using the Blueair Classic 480i smart air purifier based on the dynamic threshold calculated by this method.

[0038] Control group: The Blueair Classic 480i smart air purifier was used to purify the air and issue warnings based on fixed thresholds set by the testers (in the control group, the fixed threshold for PM2.5 was set to 75 μg / m³, and the fixed threshold for CO2 was set to 1000 ppm).

[0039] In this experiment, the environment was chosen as a square laboratory (60m). 2 ), and isolated into two areas of equal size (30 m 2The experimental group and the control group worked in two areas for 8 hours. The main pollutants monitored were PM2.5 and CO2. After 8 hours, the average concentration of PM2.5, the average concentration of CO2 (ppm), and the effective warning rate were recorded in the two areas. The effective warning rate = number of effective warnings / total number of warnings.

[0040] II. Experimental Results Table 1 Comparison of Purification Effects Table 1 shows that the average PM2.5 concentration and CO2 concentration in the experimental group were lower than those in the control group. Specifically, the average PM2.5 concentration in the experimental group was about 37% lower than that in the control group, and the average CO2 concentration was about 23% lower. The reason for this is that the experimental group was able to lower the dynamic threshold in advance before the pollutant concentration increased, thereby treating the pollutants in advance and maintaining them at a lower level.

[0041] Table 1 shows that the effective warning rate of the experimental group was higher than that of the control group. The reason for this is that the experimental group adjusted the dynamic threshold according to the changes of pollutants in the environment, which improved the alarm sensitivity.

[0042] Example 2:

[0043] Unlike the embodiments described above, as Figure 2 As shown, an air purification control system based on neural networks includes an image acquisition module, an environmental recognition module, a human recognition module, a dynamic analysis module, and a dynamic control module, all of which are interconnected.

[0044] The specific functions of each module are as follows: The image acquisition module includes an image recording device (a camera is selected in this embodiment); the image acquisition module is used to acquire real-time images of the current environment using the image recording device, generate live images, and transmit the live images to the environment recognition module and the person recognition module in real time.

[0045] Specifically, in this embodiment, the cameras are mainly installed in the corners of the room and the center of the ceiling to reduce blind spots. The cameras continuously capture video streams at a rate of 15-30 frames per second. The environment recognition module performs preliminary preprocessing on the video stream, including frame extraction (converting the video stream into continuous image frames), image compression (to reduce transmission bandwidth), and timestamp marking, to obtain live video. The image acquisition module then transmits the processed live video (i.e., the continuous sequence of image frames) to the environment recognition module and the person recognition module in real time via a local area network (such as Wi-Fi) or bus.

[0046] The environment recognition module receives real-time images acquired by the image acquisition module, extracts environmental features (including environmental contours and environmental landmarks) from the real-time images, constructs an environment recognition model based on a neural network algorithm, analyzes the environmental features using the environment recognition model, obtains the environmental type and environmental extent, and transmits the obtained environmental type and environmental extent to the dynamic analysis module.

[0047] Specifically, the environment recognition model uses an edge detection algorithm (in this embodiment, the Canny algorithm is used) to extract the physical boundaries of the room from the live image, including the intersection lines of walls, floors, and ceilings, thereby calculating the approximate area and three-dimensional structure of the room, thus obtaining the environmental range. It then uses an object detection algorithm (in this embodiment, the YOLO algorithm is used) to identify key objects in the image, such as desks, computers, conference tables, sofas, beds, stoves, range hoods, lab benches, and printers, thereby analyzing the environment type. For example, if the environment recognition model identifies multiple desks, multiple computers, and printers, it determines the environment type to be an office.

[0048] The person recognition module receives live images captured by the image acquisition module, extracts person images from the live images, constructs a person recognition model based on computer vision and deep learning algorithms, analyzes person features (including person outline, person posture, and carried items) using the person recognition model, obtains the average pedestrian flow, current pedestrian flow, and person behavior in the environment, predicts the type and concentration of pollutants generated by person behavior, and generates behavior impact parameters; and transmits the average pedestrian flow, current pedestrian flow, and behavior impact parameters to the dynamic analysis module.

[0049] Specifically, firstly, a person detection model (built based on computer vision and deep learning algorithms) is used to process each frame of the image, locating the positions and bounding boxes of all people. Then, these bounding box regions are cropped from the original image to obtain independent person images. A pose estimation algorithm (OpenPose is used in this embodiment) is used to analyze the position and motion trajectory of the skeletal key points (head, hands, feet, and torso). An object detection algorithm is used to perform detailed identification of items in the person's hands (such as cigarettes, mobile phones, books, documents, coffee cups, spray bottles, and food). For example, if the person detection model identifies a person repeatedly bringing their hand close to their mouth and also detects a cigarette in their hand, the behavior will be classified as "smoking," and the predicted pollutant will be PM2.5. The PM2.5 concentration in the environment is expected to increase from 20 μg / m³ to 120 μg / m³ (PM2.5 concentration = PM2.5 content / spatial volume).

[0050] The dynamic analysis module receives the environmental type and scope from the environmental identification module, analyzes typical pollutants within the environmental type, maps corresponding monitoring and control equipment, generates an equipment installation report, and transmits the report to the dynamic control module. It also sets basic thresholds for each pollutant based on the environmental scope. Furthermore, it receives current pedestrian traffic and behavioral impact parameters from the people and people identification module, sets dynamic adjustment coefficients based on average and current pedestrian traffic, calculates the current dynamic thresholds of the environment in real time based on the dynamic adjustment coefficients and behavioral impact parameters, and transmits these dynamic thresholds to the dynamic control module in real time.

[0051] Specifically, different types and sizes of environments have varying tolerances for various pollutants. For example, the basic CO2 threshold for a 100-cubic-meter bedroom can be set at 1000 ppm; while for a 100-cubic-meter conference room, the basic CO2 threshold can be set at 800 ppm, because conference rooms have a higher density of people than bedrooms, leading to faster CO2 accumulation, thus requiring a lower basic threshold; furthermore, for an 80-cubic-meter conference room, the basic CO2 threshold can be set at 600 ppm, because this smaller conference room can hold a smaller total amount of CO2, thus requiring stricter standards to ensure air purification efficiency.

[0052] The formula for calculating the dynamic threshold is as follows: △T=T0·KP (1).

[0053] Where △T is the dynamic threshold, T0 is the basic threshold, K is the dynamic adjustment coefficient, the value of K ranges from 0.1 to 2, and P is the behavior influence parameter.

[0054] The formula for calculating the dynamic adjustment coefficient is as follows: K=N p / N N (2).

[0055] Where, N p N represents the average foot traffic over the past 7 days. N N represents the current foot traffic. When the foot traffic is 0, N represents the current foot traffic. N Take 1.

[0056] Specifically, taking the average daily foot traffic N in office A over the past 7 days as an example... p =20 people, current visitor flow N N=25 people, according to formula (2), K=0.8, which means that the dynamic threshold is reduced and the purification standard is increased at this time, so as to ensure that the air in office A can adapt to more people. Assuming that there are only 2 cleaning staff left in office A after evening, according to formula (2), K=10>2, so K is 2, which means that the dynamic threshold is increased and the purification standard is reasonably reduced to avoid over-purification and save energy. At the same time, by limiting the value of K, the dynamic threshold is avoided from being too high, resulting in the purification standard being too low, so as to ensure the rationality of the purification work. Assuming that the basic threshold of VOC in office A is set at 500ppm, at this time, two cleaning staff are spraying alcohol to disinfect the glass. The personnel recognition module will recognize the disinfection behavior and estimate the change in VOC concentration. It is expected that the VOC concentration will rise from 30ppm to 130ppm. The influence parameter P=100. The dynamic analysis module will calculate according to formula (1) and obtain the dynamic threshold △T=900ppm.

[0057] The dynamic control module includes several monitoring and control devices. It receives equipment installation reports and dynamic thresholds from the dynamic analysis module and transmits them to the user terminal, allowing the user to install monitoring equipment (including several gas concentration detectors) and control devices (including several air filters, exhaust fans, air inlet ducts, and exhaust ducts) in the environment based on the monitoring equipment installation reports. It uses the monitoring equipment to monitor the concentration of various pollutants in the environment in real time, generating real-time concentrations and comparing them with dynamic thresholds. If the real-time concentration exceeds the dynamic threshold, the dynamic control module controls the operation of the corresponding control devices to regulate the air in the environment. The control methods include activating air filters, adjusting exhaust fan speed, and adjusting the number of opening and closing air inlet and exhaust ducts.

[0058] When the dynamic threshold is less than or equal to the real-time concentration and less than 110% of the dynamic threshold, the dynamic control module issues a Level 1 warning (reminding the user that the pollutant concentration is slightly exceeded); when the dynamic threshold is less than or equal to the real-time concentration and less than 130% of the dynamic threshold, the dynamic control module issues a Level 2 warning (reminding the user that the pollutant concentration is moderately exceeded); when the dynamic threshold is less than or equal to the real-time concentration, the dynamic control module issues a Level 3 warning (reminding the user that the pollutant concentration is severely exceeded).

[0059] Specifically, the dynamic control module receives an equipment installation report from the dynamic analysis module. The equipment installation report lists the equipment type and its installation location. For example, it specifies that one set each of CO2 and PM2.5 sensors needs to be installed. The CO2 sensor should be installed on a wall 1.2-1.5 meters above the ground, away from doors, windows, and vents; the PM2.5 sensor should be installed in a well-ventilated area, avoiding corners. The dynamic control module clearly presents this information to the user via a user terminal (such as a mobile phone, computer, or tablet; this embodiment uses a mobile phone), guiding the user to correctly install and network the hardware, ensuring the accuracy of data collection.

[0060] Assuming the current dynamic threshold for CO2 in room B is 800 ppm, and the dynamic control module detects a real-time CO2 concentration of 820 ppm based on the CO2 sensor, then since 800 ppm < 820 ppm < 880 ppm, the module will send a Level 1 warning to the user's mobile phone, alerting them that the room's CO2 concentration is slightly above the limit (exceeding the dynamic threshold by 20 ppm). Simultaneously, the module will increase the exhaust fan speed by 10% and open an additional exhaust duct to accelerate CO2 removal. The dynamic control module then compares the dynamic threshold with the concentrations of various pollutants in the environment. Based on the comparison results, it issues different levels of warnings and executes corresponding control measures. This allows for appropriate treatment of pollution at different levels of severity, ensuring effectiveness while avoiding overtreatment, thus improving the efficiency and rationality of the treatment process.

[0061] Existing air purification methods rely excessively on fixed thresholds, lack air quality prediction, cannot dynamically adapt to environmental changes, and have low purification efficiency. In contrast to existing air purification methods, this invention can monitor the flow of people and their behavior in the environment in real time, predict the impact of these factors on air quality, generate dynamic adjustment coefficients and behavior impact parameters, and dynamically adjust the basic threshold to adapt to environmental changes and improve air purification efficiency.

[0062] This invention can analyze the environmental type and range based on real-time environmental images, allowing users to set more reasonable basic thresholds according to the environmental type and range. This enables the purification method of this solution to better purify the air in different environments, improving the adaptability of this solution to the environment and the air purification effect.

[0063] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. An air purification control method based on neural networks, characterized in that, Includes the following steps: S1, Environmental Analysis: Real-time acquisition of environmental images; analysis of environmental type and scope based on environmental images; setting basic thresholds for various pollutants based on environmental type and scope; and setting different types and levels of early warning prompts and control schemes. S2, People Analysis: Based on real-time collected environmental images, analyze the average and current pedestrian flow and people's behavior in the environment, and set dynamic adjustment coefficients based on the average and current pedestrian flow; analyze the types and concentrations of pollutants that people may generate based on their behavior, and set behavior impact parameters; S3, Dynamic Threshold Calculation: Based on the dynamic adjustment coefficient and the behavior influence parameter, the basic threshold is dynamically adjusted to obtain the dynamic threshold; S4, Dynamic Air Control: Monitors the concentration of various pollutants in the environment and compares the concentration of various pollutants with their dynamic thresholds. When the concentration of any type of pollutant exceeds its corresponding dynamic threshold, an early warning of the corresponding type and level will be issued based on the type of pollutant and the amount exceeding the threshold, and the corresponding control plan will be activated.

2. The air purification control method based on neural networks according to claim 1, characterized in that, The warning system includes Level 1, Level 2, and Level 3 warnings, with the warning level being directly proportional to the pollutant concentration.

3. A neural network-based air purification control system, based on the neural network-based air purification control method described in any one of claims 1-2, characterized in that, It includes an image acquisition module, an environment recognition module, a person recognition module, a dynamic analysis module, and a dynamic control module; The image acquisition module includes an image recording device; the image acquisition module is used to acquire real-time images of the current environment using the image recording device, generate live images, and transmit the live images to the environment recognition module and the person recognition module in real time. The environment recognition module is used to receive real-time images acquired by the image acquisition module, extract environmental features from the real-time images, construct an environment recognition model based on a neural network algorithm, analyze the environmental features using the environment recognition model, obtain the environment type and environmental range, and transmit the obtained environment type and environmental range to the dynamic analysis module. The person recognition module receives live images captured by the image acquisition module, extracts person images from the live images, constructs a person recognition model based on computer vision and deep learning algorithms, analyzes person characteristics using the person recognition model, obtains average pedestrian traffic, current pedestrian traffic and person behavior in the environment, predicts the type and concentration of pollutants generated by person behavior, generates behavior impact parameters, and transmits the average pedestrian traffic, current pedestrian traffic and behavior impact parameters to the dynamic analysis module. The dynamic analysis module receives the environmental type and environmental range obtained by the environmental identification module, analyzes the typical pollutants in the environmental type according to the environmental type, maps the corresponding monitoring and control equipment, generates an equipment installation report, and transmits the equipment installation report to the dynamic control module. Used to set basic thresholds for each pollutant based on the environmental scope; It is used to receive the current traffic flow and behavior impact parameters from the person recognition module, set the dynamic adjustment coefficient based on the average traffic flow and the current traffic flow, calculate the current dynamic threshold of the environment in real time based on the dynamic adjustment coefficient and behavior impact parameters, and transmit the dynamic threshold to the dynamic control module in real time. The dynamic control module includes several monitoring and control devices. It receives equipment installation reports and dynamic thresholds from the dynamic analysis module and transmits them to the user terminal, allowing the user to install monitoring and control devices in the environment based on the monitoring equipment installation reports. It also uses the monitoring devices to monitor the concentration of various pollutants in the environment in real time, generate real-time concentrations, and compares the real-time concentrations with the dynamic thresholds. If the real-time concentration is greater than the dynamic threshold, the dynamic control module controls the operation of the corresponding control devices.

4. The air purification control system based on a neural network according to claim 3, characterized in that, When the dynamic threshold ≤ real-time concentration < 110% of the dynamic threshold, the dynamic control module is used to issue a Level 1 warning; when the dynamic threshold ≤ real-time concentration < 130% of the dynamic threshold, the dynamic control module is used to issue a Level 2 warning; when the dynamic threshold ≤ real-time concentration, the dynamic control module is used to issue a Level 3 warning.

5. The air purification control system based on a neural network according to claim 4, characterized in that, The formula for calculating the dynamic threshold is as follows: △T=T0·KP (1; Where △T is the dynamic threshold, T0 is the basic threshold, K is the dynamic adjustment coefficient, the value of K ranges from 0.1 to 2, and P is the behavior influence parameter.

6. The air purification control system based on a neural network according to claim 5, characterized in that, The formula for calculating the dynamic adjustment coefficient is as follows: K=N p / N N (2); Where, N p N represents the average foot traffic over the past 7 days. N N represents the current foot traffic. When the foot traffic is 0, N represents the current foot traffic. N Take 1.

7. The air purification control system based on a neural network according to claim 6, characterized in that, Environmental features include environmental outlines and environmental landmarks, while human features include human outlines, human postures, and objects carried.

8. The air purification control system based on a neural network according to claim 7, characterized in that, The monitoring equipment includes several gas concentration detectors.

9. The air purification control system based on a neural network according to claim 8, characterized in that, The control equipment includes several air filters, exhaust fans, air intake ducts, and air exhaust ducts.

10. The air purification control system based on a neural network according to claim 9, characterized in that, The control methods of the control scheme include activating the air filter, adjusting the exhaust fan speed, and adjusting the number of opening and closing air intake and exhaust ducts.

Images