Fresh air system of duty platform and fresh air control method

Abstract

The invention relates to the technical field of air conditioning, in particular to a fresh air system of an on-duty platform and a fresh air control method. The system comprises an environment parameter collecting module, a personnel activity sensing module, a control module and a fresh air adjusting module. The environment parameter acquisition module monitors the space environment quality index in the on-duty platform, the personnel activity sensing module obtains the personnel existence state and activity intensity information, and the control module dynamically calculates the fresh air demand value through a built-in control algorithm according to the data and generates a control signal. The fresh air adjusting module responds to the control signal to adjust the fresh air inlet air volume. The control method comprises the steps of collecting environment parameters and personnel activity data, executing multivariable analysis to calculate the target fresh air volume, driving the fresh air execution unit to adjust the air volume and the like. The dynamic demand model is constructed through multi-source data fusion, the model parameters are optimized through an online learning mechanism, and the environment control precision and the energy utilization efficiency are improved through self-adaptive fresh air regulation and control according to the actual use condition of the on-duty platform.

Description

Technical Field

[0001] This invention relates to the field of air conditioning technology, and more specifically, to a fresh air system and fresh air control method for a duty platform. Background Technology

[0002] As a special workplace, the internal environmental quality of duty platforms directly affects the work efficiency and health of personnel on duty. Maintaining a suitable air environment is crucial for ensuring the physiological comfort and psychological well-being of personnel on long-term duty. Traditional environmental control systems often struggle to adapt to the unique working patterns and personnel activity characteristics of duty platforms, necessitating more intelligent environmental control solutions.

[0003] Existing environmental control systems for duty platforms suffer from the following shortcomings: Most systems employ fixed threshold control strategies, failing to dynamically adjust based on actual personnel activity, leading to a mismatch between fresh air supply and actual demand. Control methods based on single environmental parameters struggle to comprehensively reflect spatial environmental quality; for example, controlling ventilation solely based on carbon dioxide concentration ignores the combined effects of other influencing factors such as temperature and humidity. Traditional systems lack the ability to perceive the intensity of personnel activity, failing to differentiate environmental requirements between static duty and high-intensity activity. Most control algorithms use preset parameters, lacking adaptive learning capabilities and unable to optimize control strategies based on long-term operational data. There is a lack of effective data fusion mechanisms between system modules; environmental parameters and personnel activity information are often processed independently, hindering comprehensive judgment. Furthermore, existing technologies often employ periodic, fixed-frequency data acquisition and processing, unable to dynamically adjust monitoring frequency according to the urgency of environmental changes, resulting in delayed responses in emergencies. These deficiencies frequently lead to contradictory situations in practical applications: excessive ventilation causing energy waste or insufficient ventilation impacting environmental quality.

[0004] Therefore, a fresh air system and fresh air control method for duty platforms are proposed to address the above problems. The technical issues to be solved include: how to accurately calculate the fresh air demand based on the real-time changes in environmental parameters and personnel activities within the duty platform, how to establish dynamic control strategies that adapt to different work scenarios, and how to continuously optimize the system control performance through a self-learning mechanism. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a fresh air system and a fresh air control method for a duty platform to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a fresh air system for a duty platform, comprising an environmental parameter acquisition module, a personnel activity sensing module, a control module, and a fresh air adjustment module. The environmental parameter acquisition module is used to continuously monitor the spatial environmental quality indicators inside the duty platform. The personnel activity sensing module is used to acquire information on the presence status and activity intensity of personnel within the duty platform in real time. The control module is communicatively connected to the environmental parameter acquisition module and the personnel activity sensing module, and is configured to dynamically calculate the fresh air demand value based on the spatial environmental quality indicators and the personnel presence status and activity intensity information using a built-in control algorithm, and generate a control signal based on the fresh air demand value. The fresh air adjustment module responds to the control signal by continuously adjusting or adjusting the fresh air inlet air volume in stages. The control algorithm includes multi-source data fusion of the spatial environmental quality indicators and the personnel presence status and activity intensity information to construct a dynamic demand model, and real-time updating of model parameters through an online learning mechanism, thereby adjusting the output frequency and amplitude of the control signal according to the fusion result and model output. The online learning mechanism uses a recursive optimization method to adjust the model weights based on real-time data streams to capture the nonlinear changing trends of the environment and personnel activities.

[0007] Preferably, the space environment quality indicators include at least two of the following: carbon dioxide concentration, temperature, humidity, volatile organic compound concentration, and fine particulate matter concentration. The environmental parameter acquisition module acquires these indicators at fixed time intervals and sets different acquisition priorities and data filtering methods for different indicators to ensure the comprehensiveness and real-time performance of the monitoring. The data filtering methods include moving average or low-pass filtering strategies.

[0008] Preferably, the personnel presence status and activity intensity information includes at least one of the following: number of personnel, frequency of personnel movement, duration of personnel stay, and personnel distribution density. The personnel activity sensing module acquires this information through multi-source data fusion technology, which involves comprehensive analysis of infrared signals, sound waveforms, and motion patterns, and uses time series analysis to distinguish between short-term fluctuations and long-term trends, so as to improve the accuracy and robustness of personnel activity detection.

[0009] Preferably, the dynamic demand model combines the spatial environment quality indicators and personnel presence status and activity intensity information through linear or nonlinear functions, and outputs a comprehensive score. The control module adjusts the control signal based on the comparison result of the comprehensive score and the dynamic threshold. The dynamic threshold is periodically updated based on the historical operation data of the duty platform and external environmental conditions through sliding window technology, and the update frequency of the dynamic threshold is adjusted in the range of once per minute to once per hour.

[0010] Preferably, the online learning mechanism uses gradient descent or least squares to optimize the parameters of the dynamic demand model, and the model parameters are adaptively adjusted according to real-time monitoring data through an error feedback loop. The error feedback loop includes calculating the deviation between the predicted value and the actual value, and adjusting the learning rate according to the magnitude of the deviation. The learning rate is set in the range of 0.01 to 0.5.

[0011] A method for controlling fresh air in a duty platform includes the following steps: S1 collects various environmental parameters inside the duty platform in real time through the environmental monitoring unit; S2 continuously acquires personnel activity status data within the duty platform through the personnel detection unit; S3 uses the processing unit to run a control program, which performs multivariate comprehensive analysis based on the various environmental parameters and personnel activity status data to calculate the target fresh air volume and generate corresponding control commands; S4 drives the fresh air execution unit to adjust the fresh air flow according to the control command; The multivariate comprehensive analysis includes inputting various environmental parameters and personnel activity status data into a dynamic demand model. The model optimizes its parameters in real time through an online learning mechanism and outputs a comprehensive demand index. The processing unit queries a predefined control table based on the comprehensive demand index to generate control instructions, and adjusts the parameters of the control instructions based on real-time feedback. The online learning mechanism adopts an iterative optimization process to update the model coefficients based on data flow.

[0012] Preferably, the multiple environmental parameters include at least three of the following: carbon dioxide concentration, temperature, humidity, volatile organic compound concentration, and fine particulate matter concentration. When collecting these environmental parameters, a multi-sensor data fusion method is used to eliminate the error of a single sensor and improve data reliability. The multi-sensor data fusion method includes a Kalman filter or a weighted averaging strategy, and the data acquisition frequency is set in the range of once per second to once every 5 minutes.

[0013] Preferably, the personnel activity status data includes at least two of the following: number of personnel, personnel movement speed, personnel dwell time, and personnel activity intensity. When acquiring this data, time series analysis and pattern recognition technology are combined to distinguish between short-term fluctuations and long-term trends, thereby more accurately reflecting the personnel dynamics within the duty platform. The pattern recognition technology includes clustering analysis or classification algorithms.

[0014] Preferably, the dynamic demand model combines various environmental parameters and personnel activity status data through linear or nonlinear functions and outputs a comprehensive score. The processing unit queries a predefined control table based on the comprehensive score to generate control instructions. The control table is trained based on a large amount of experimental data and covers fresh air volume adjustment strategies under different scenarios. The control table is updated regularly through an online learning mechanism.

[0015] Preferably, the online learning mechanism uses recursive least squares or stochastic gradient descent to optimize the parameters of the dynamic demand model, and the model parameters are adaptively adjusted through an error feedback loop based on real-time monitoring data. The error feedback loop includes calculating the deviation between the predicted value and the actual value, and adjusting the learning rate according to the magnitude of the deviation. The learning rate is set in the range of 0.01 to 0.5. Meanwhile, the model optimization process is executed after each batch of data is processed.

[0016] The technical effects and advantages of this invention are as follows: Compared to existing technologies, this invention constructs a dynamic demand model through multi-source data fusion technology. It comprehensively analyzes environmental parameters and personnel activity information, employing linear or nonlinear functions to combine multiple input variables and outputs a comprehensive score in real time as the basis for control. The system continuously optimizes model parameters through an online learning mechanism and uses a recursive optimization method to adjust weight coefficients based on real-time data streams, accurately capturing the nonlinear changes in the environment and personnel activities. This design enables the system to automatically adjust control strategies according to the actual usage of the duty platform, ensuring environmental quality stability, improving energy efficiency, and avoiding the response lag or over-adjustment problems caused by fixed threshold control.

[0017] Compared to existing technologies, this invention establishes a comprehensive online learning mechanism, employing gradient descent or least squares methods to continuously optimize the parameters of the dynamic demand model. Combined with error feedback, it cyclically adjusts the learning rate in real time, adaptively setting the optimization speed within the range of 0.01 to 0.5. This mechanism dynamically corrects model parameters by calculating the deviation between predicted and actual values, enabling the system to gradually adapt to the unique usage patterns of different duty platforms. This design gives the control system continuous improvement capabilities, constantly optimizing control accuracy over time, effectively overcoming the limitations of traditional fixed-parameter systems that struggle to adapt to changing environments, and enhancing the system's adaptability and stability in different seasons and usage scenarios.

[0018] Compared to existing technologies, this invention introduces a multi-sensor data fusion method to process environmental parameters. It integrates various environmental monitoring data using Kalman filtering or weighted averaging strategies, effectively eliminating measurement errors from single sensors. The system collects environmental parameters at an adjustable frequency of once per second to once every five minutes, combining time series analysis and pattern recognition techniques to process personnel activity data. Cluster analysis or classification algorithms are used to distinguish between short-term fluctuations and long-term trends. This multi-level data processing approach significantly improves the accuracy and reliability of the system's perception, providing a high-quality data foundation for control decisions, avoiding control errors caused by single-point data anomalies, and enhancing the system's robustness in complex environments. Attached Figure Description

[0019] Figure 1 This is a diagram of the overall system architecture of the present invention.

[0020] Figure 2 This is a flowchart of the data acquisition and processing process of the present invention.

[0021] Figure 3 This is the control algorithm and online learning diagram of the present invention.

[0022] Figure 4 This is a flowchart of the fresh air control method of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 1 As attached Figures 1 to 4 The diagram illustrates a fresh air system and control method for a duty platform. This system is built on a distributed architecture and comprises four main functional modules forming a complete technical system. The environmental parameter acquisition module comprehensively monitors various air quality indicators within the duty environment, employing multi-sensor fusion technology to ensure data integrity. The personnel activity sensing module tracks the distribution and activity status of personnel within the platform in real time through a heterogeneous sensor network, providing behavioral data for system decision-making. The control module, as the system's intelligent hub, integrates advanced data analysis and machine learning capabilities, processing multi-source information and generating optimized control strategies. The fresh air adjustment module, as the final execution unit, precisely adjusts the air supply parameters according to control commands, achieving proactive regulation of environmental conditions.

[0025] Furthermore, in terms of environmental parameter collection, the system has established a multi-dimensional environmental quality monitoring system. The monitoring indicators cover key parameters such as carbon dioxide concentration, temperature, humidity, volatile organic compound concentration, and fine particulate matter concentration, ensuring a comprehensive understanding of the environmental status.

[0026] In actual deployment, the system adopts a hierarchical sensor network architecture, with the master node coordinating the synchronous operation of multiple sub-nodes.

[0027] Each monitoring node is equipped with a high-precision sensor array. The carbon dioxide sensor uses non-dispersive infrared technology, with a measurement range of 0-5000 ppm and a resolution of 1 ppm. The temperature and humidity sensors use capacitive measurement principles, with a temperature accuracy of ±0.3℃ and a humidity accuracy of ±2%RH. The VOC sensor uses metal-oxide-semiconductor technology, with a detection range of 0-5000 ppb. The PM2.5 sensor uses laser scattering principles, with a detection range of 0-1000 μg / m³. 3 .

[0028] These sensors operate according to an intelligent sampling strategy, with a basic sampling frequency set every 30 seconds. When a sudden change in parameters is detected, they automatically switch to a high-frequency sampling mode every 5 seconds. The collected raw data undergoes a three-stage processing flow: first, outliers are removed using median filtering; second, the data is smoothed using a moving average algorithm; and finally, high-frequency noise is eliminated using a Butterworth low-pass filter.

[0029] For parameters with a coupling relationship, such as temperature and humidity, the system establishes a compensation model based on multiple regression to effectively eliminate measurement errors caused by cross-sensitivity. This refined data acquisition and processing mechanism provides the system with accurate and reliable environmental status information.

[0030] Furthermore, in terms of personnel activity perception, the system constructs a multimodal fusion intelligent sensing network. The perceived information includes key behavioral parameters such as the number of people, the frequency of personnel movement, the duration of personnel stay, and the density of personnel distribution.

[0031] The system deploys three types of sensing terminals: an infrared thermal imaging sensor array mounted on the ceiling to capture personnel distribution using 64×48 resolution thermal images; an acoustic sensor array positioned in the corners, employing an 8-microphone array to collect audio signals; and a millimeter-wave radar sensor embedded in the wall to detect personnel movement using 77GHz electromagnetic waves. These sensing terminals coordinate their operation through a time synchronization protocol to ensure data temporal consistency. After feature extraction, the raw sensing data is input into a data fusion algorithm based on a deep neural network.

[0032] This network employs an encoder-decoder structure, with the encoder processing data from different modalities and the decoder performing feature fusion and personnel state reconstruction. The system uses a sliding window mechanism to process continuous data streams, with the window size dynamically adjusted according to the scene; it is typically set to 10 minutes, automatically shortening to 5 minutes during peak hours. Through time-frequency analysis of the data within the window, the system can accurately distinguish between instantaneous personnel movement and continuous personnel presence. This multimodal fusion sensing technology effectively overcomes the limitations of single sensors in complex environments.

[0033] Furthermore, regarding the core control algorithm, the system adopts an intelligent decision-making mechanism based on a dynamic demand model. This model transforms multi-source input data into a unified evaluation index through nonlinear mapping. Specifically, the model first normalizes each input parameter, converting it to a dimensionless range of [0,1]. The normalization function is designed according to the parameter characteristics: piecewise linear normalization is used for carbon dioxide concentration, with the inflection point set at 1000 ppm; S-shaped function normalization is used for temperature, with the center point set at 25℃; and logarithmic function normalization is used for the number of personnel to adapt to the diminishing marginal returns of increasing personnel. The dynamic demand model is mathematically expressed as a weighted power function: in This indicates the overall score. Indicates the first One environmental parameter, Indicates the first Individual activity parameters, and These represent the corresponding weight coefficients. and A normalization function specific to the parameters is used. Weight coefficients are determined using the analytic hierarchy process (AHP) and corrected using entropy weighting to ensure that the weight allocation reflects both expert experience and data characteristics. This model structure effectively captures the interactions between parameters, particularly the impact of human activity on air quality in high-temperature and high-humidity environments.

[0034] Furthermore, the system employs an online learning mechanism based on reinforcement learning to continuously optimize model parameters. This mechanism establishes an intelligent optimization system by constructing a complete closed loop of environmental state-control action-environment response. The system state space is defined as the joint feature vector of environmental parameters and personnel activity parameters, the action space is defined as the set of control commands for the fresh air equipment, and the reward function comprehensively considers multiple objectives such as environmental comfort, energy efficiency, and equipment lifespan. Parameter updates utilize an actor-critic algorithm framework, where the actor network generates control strategies, and the critic network evaluates the value of the strategies. The network parameter update formula is: in and These represent the parameters of the actor network and the critic network, respectively. and For learning rate, This is a discount factor. The learning rate is dynamically adjusted based on the model's convergence state and optimized using an adaptive moment estimation algorithm. This reinforcement learning framework enables the system to continuously accumulate experience during actual operation, gradually improving its control performance.

[0035] Furthermore, regarding the fresh air control method, the system has established a standardized closed-loop control process. The control process consists of four ordered steps: the environmental parameter acquisition step acquires multi-dimensional environmental data through a distributed sensor network to ensure comprehensive and real-time monitoring; the personnel status perception step identifies personnel distribution and activity characteristics through multi-modal fusion technology, providing behavioral basis for demand calculation; the intelligent decision-making step calculates the optimal control parameters based on a dynamic demand model and generates specific control commands; and the execution and regulation step drives the fresh air equipment to adjust its operating status, achieving precise regulation of environmental conditions. Each step includes a quality inspection process to ensure the controllability and reliability of data processing.

[0036] Furthermore, in the environmental parameter acquisition stage, the system employs a data fusion algorithm based on distributed Kalman filtering. This algorithm describes the evolution of environmental parameters using a state-space model, fusing observations from multiple sensors into an optimal estimate. The system state equation is defined as: The observation equation is defined as: in Represents the system state vector. Represents the observation vector. Here is the state transition matrix. For the observation matrix, and These are process noise and observation noise, respectively. The algorithm iterates through two stages: prediction and update. In the prediction stage, the current state is estimated based on the system model, and in the update stage, the estimation result is corrected using sensor observations. This distributed fusion architecture effectively improves the robustness of the system, maintaining basic environmental monitoring capabilities even if some sensors fail.

[0037] Furthermore, in terms of personnel activity status identification, the system employs a joint analysis method based on spectral clustering and hidden Markov models. The spectral clustering algorithm divides the duty platform into multiple regions with distinct functional characteristics, and its similarity matrix is ​​constructed based on the spatiotemporal patterns of personnel flow. The Laplace matrix eigenvalue decomposition is expressed as: in This is a similarity matrix. The Laplace matrix is ​​used as the degree matrix. By performing eigenvalue decomposition on the Laplace matrix, the original data is mapped to a low-dimensional feature space, and then clustering is performed using the k-means algorithm. Hidden Markov Models (HMMs) are used to describe the state transition patterns of human activities within each region. The probability distribution of the observed sequence is calculated using a forward-backward algorithm, and the most probable state sequence is decoded using the Viterbi algorithm. This joint analysis method can accurately capture the spatiotemporal patterns of human activities, providing refined input information for differentiated fresh air control.

[0038] Furthermore, in terms of dynamic demand modeling, the system employs an intelligent reasoning mechanism based on fuzzy cognitive graphs. This mechanism abstracts environmental parameters and personnel activity parameters into concept nodes, and represents the causal relationships between concepts through directed edges. The activation value of a concept node continuously varies within the range [0,1], representing the degree of contribution of that concept to the system state. The strength of the causal relationships between concepts is weighted using a weight matrix. It means that among them Representing concepts On the concept The degree of influence. The evolution of the system state is achieved through iterative calculation: in Indicates the first The concept activation value vector of the next iteration It uses an sigmoid activation function. This fuzzy cognitive graph model can well express the complex nonlinear relationships between system parameters, especially the dynamic coupling characteristics between environmental parameters and human activities.

[0039] Furthermore, regarding parameter optimization, the system employs an online learning framework based on Bayesian inference. This framework treats model parameters as random variables and updates the posterior distribution of the parameters using Bayes' theorem: in Indicates model parameters, Represents observation data, For the prior distribution, Let be the likelihood function. Let be the posterior distribution. Since the posterior distribution is difficult to calculate directly, the system uses variational inference to approximate it, approximating the true posterior by optimizing the lower bound of evidence. The variational lower bound is defined as: in It uses a variational distribution. This Bayesian learning framework enables the system to quantify the uncertainty of parameter estimation, achieving a good balance between exploration and exploitation.

[0040] To illustrate the system's working mechanism, consider a typical application scenario: environmental control at a border post during the low-temperature period in winter.

[0041] At 6:00 AM, the environmental monitoring unit detected an indoor temperature of only 16°C and a carbon dioxide concentration of 800 ppm. The personnel detection unit identified two personnel on duty at the outpost beginning their routine patrol, with the activity intensity level transitioning from low to medium. The system input the environmental parameters and personnel activity data into the dynamic demand model, and through multi-dimensional calculations, obtained a comprehensive score of 65 points.

[0042] After consulting the expert knowledge base, the system generates tiered control instructions: First, start the fresh air preheating mode and set the supply air temperature to 20℃; then, gradually increase the fresh air volume according to the intensity of personnel activity, from the initial 30% of the rated air volume to 60%.

[0043] Upon receiving the instruction, the fresh air execution unit sequentially adjusts the heat recovery device, electric heater, and variable frequency fan to achieve a smooth transition. The system continuously monitors changes in environmental parameters and reassesses the control strategy every 10 minutes. Anticipating an increase in personnel density during the 8:00 AM shift change, it proactively increases the fresh air volume to 75%, effectively preventing an increase in carbon dioxide concentration.

[0044] Throughout the entire control process, the system records environmental response characteristics through an online learning mechanism, optimizes subsequent control parameters, and forms a virtuous cycle.

[0045] In terms of system architecture design, the communication network adopts a hybrid networking of industrial Ethernet and wireless sensor network. Key data is transmitted via wired transmission to ensure reliability, while mobile nodes are connected wirelessly to improve flexibility.

[0046] Data processing adopts a collaborative architecture of edge computing and cloud computing, with real-time control tasks completed on edge nodes and long-term optimization tasks executed on cloud servers.

[0047] The system was deployed with full consideration of the special environment of the duty platform. All equipment meets the requirements for dust and moisture protection, and electromagnetic compatibility meets military standards.

[0048] This new air control system, based on multi-source information fusion and adaptive learning, achieves proactive protection and efficient management of the internal environmental quality of the duty platform through refined environmental perception, intelligent decision support, and precise execution control.

[0049] The system demonstrates good adaptability and robustness in actual operation, and can automatically adjust control strategies according to environmental changes and personnel activities, optimizing energy efficiency while ensuring environmental comfort, and creating a better working environment for on-duty personnel.

[0050] Finally, the following points should be noted: First, in the description of this application, it should be noted that, unless otherwise specified and limited, the terms "installation", "connection", and "linkage" should be interpreted broadly, and can be mechanical or electrical connections, or internal connections between two components, or direct connections. "Up", "down", "left", "right", etc. are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may change. Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A fresh air system for a duty platform, characterized in that... The system includes an environmental parameter acquisition module, a personnel activity sensing module, a control module, and a fresh air regulation module. The environmental parameter acquisition module continuously monitors the spatial environmental quality indicators within the duty platform. The personnel activity sensing module acquires real-time information on the presence status and activity intensity of personnel within the duty platform. The control module is communicatively connected to the environmental parameter acquisition module and the personnel activity sensing module and is configured to dynamically calculate the fresh air demand value based on the spatial environmental quality indicators and the personnel presence status and activity intensity information using a built-in control algorithm, and generate a control signal based on the fresh air demand value. The fresh air regulation module responds to the control signal by continuously or incrementally adjusting the fresh air inlet air volume. The control algorithm involves fusing multi-source data from the spatial environmental quality indicators and the personnel presence status and activity intensity information to construct a dynamic demand model, and updating the model parameters in real-time through an online learning mechanism. This allows the output frequency and amplitude of the control signal to be adjusted based on the fusion results and model output. The online learning mechanism uses a recursive optimization method to adjust the model weights based on real-time data streams to capture the nonlinear changing trends of the environment and personnel activities.

2. The fresh air system for a duty platform according to claim 1, characterized in that... The space environment quality indicators include at least two of the following: carbon dioxide concentration, temperature, humidity, volatile organic compound concentration, and fine particulate matter concentration. The environmental parameter acquisition module collects these indicators at fixed time intervals and sets different acquisition priorities and data filtering methods for different indicators to ensure the comprehensiveness and real-time performance of the monitoring. The data filtering methods include moving average or low-pass filtering strategies.

3. The fresh air system for a duty platform according to claim 1, characterized in that... The information on the presence and activity intensity of personnel includes at least one of the following: number of personnel, frequency of personnel movement, duration of personnel stay, and personnel distribution density. The personnel activity sensing module acquires this information through multi-source data fusion technology, which involves the comprehensive analysis of infrared signals, sound waveforms, and motion patterns, and uses time series analysis to distinguish between short-term fluctuations and long-term trends, so as to improve the accuracy and robustness of personnel activity detection.

4. The fresh air system for a duty platform according to claim 1, characterized in that... The dynamic demand model combines the spatial environment quality indicators and personnel presence and activity intensity information through linear or nonlinear functions, and outputs a comprehensive score. The control module adjusts the control signal based on the comparison result of the comprehensive score and the dynamic threshold. The dynamic threshold is periodically updated based on the historical operation data of the duty platform and external environmental conditions through sliding window technology, and the update frequency of the dynamic threshold is adjusted in the range of once per minute to once per hour.

5. The fresh air system for a duty platform according to claim 4, characterized in that... The online learning mechanism uses gradient descent or least squares to optimize the parameters of the dynamic demand model, and the model parameters are adaptively adjusted through an error feedback loop based on real-time monitoring data. The error feedback loop includes calculating the deviation between the predicted value and the actual value, and adjusting the learning rate according to the magnitude of the deviation. The learning rate is set in the range of 0.01 to 0.

5.

6. A method for controlling fresh air in a duty platform, characterized in that... Includes the following steps: S1 collects various environmental parameters inside the duty platform in real time through the environmental monitoring unit; S2 continuously acquires personnel activity status data within the duty platform through the personnel detection unit; S3 uses the processing unit to run a control program, which performs multivariate comprehensive analysis based on the various environmental parameters and personnel activity status data to calculate the target fresh air volume and generate corresponding control commands; S4 drives the fresh air execution unit to adjust the fresh air flow according to the control command; The multivariate comprehensive analysis includes inputting various environmental parameters and personnel activity status data into a dynamic demand model. The model optimizes its parameters in real time through an online learning mechanism and outputs a comprehensive demand index. The processing unit queries a predefined control table based on the comprehensive demand index to generate control instructions, and adjusts the parameters of the control instructions based on real-time feedback. The online learning mechanism adopts an iterative optimization process to update the model coefficients based on data flow.

7. The fresh air control method for a duty platform according to claim 6, characterized in that... The various environmental parameters include at least three of the following: carbon dioxide concentration, temperature, humidity, volatile organic compound concentration, and fine particulate matter concentration. When collecting these environmental parameters, a multi-sensor data fusion method is used to eliminate the error of a single sensor and improve data reliability. The multi-sensor data fusion method includes Kalman filtering or weighted averaging strategies, and the data acquisition frequency is set in the range of once per second to once every 5 minutes.

8. The fresh air control method for a duty platform according to claim 6, characterized in that... The personnel activity status data includes at least two of the following: number of personnel, personnel movement speed, personnel dwell time, and personnel activity intensity. When acquiring this data, time series analysis and pattern recognition technology are combined to distinguish between short-term fluctuations and long-term trends, thereby more accurately reflecting the personnel dynamics within the duty platform. The pattern recognition technology includes clustering analysis or classification algorithms.

9. The fresh air control method for a duty platform according to claim 6, characterized in that... The dynamic demand model combines various environmental parameters and personnel activity status data through linear or nonlinear functions and outputs a comprehensive score. The processing unit queries a predefined control table based on the comprehensive score to generate control instructions. The control table is trained based on a large amount of experimental data and covers fresh air volume adjustment strategies under different scenarios. The control table is updated regularly through an online learning mechanism.

10. The fresh air control method for a duty platform according to claim 9, characterized in that... The online learning mechanism uses recursive least squares or stochastic gradient descent to optimize the parameters of the dynamic demand model. The model parameters are adaptively adjusted through an error feedback loop based on real-time monitoring data. The error feedback loop includes calculating the deviation between the predicted value and the actual value and adjusting the learning rate according to the magnitude of the deviation. The learning rate is set in the range of 0.01 to 0.

5. The model optimization process is executed after each batch of data is processed.

Images