Intelligent building sampling detection control system based on Internet of Things sensing

Abstract

The invention relates to an intelligent building sampling detection control system based on Internet of Things sensing, which belongs to the technical field of building environment control and comprises an environment state sensing unit, a predictive analysis unit, a flow field preprocessing unit, a path planning unit and a cooperative control unit. The environment state sensing unit is used for collecting environment state data of a target monitoring area to generate a real-time data vector; the predictive analysis unit is used for receiving the real-time data vector and performing diffusion trajectory deduction according to a preset state space normal form so as to generate a predictive state vector; the flow field preprocessing unit is used for responding to the prediction state vector and carrying out vortex kinetic energy deconstruction processing on the thermal plume so as to generate a weakened flow field environment, precious decision-making advance is won for follow-up accurate removal, and due to the fact that a pre-trained model is adopted, the real-time performance of prediction is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of building environment control, specifically to a smart building sampling and detection control system based on Internet of Things (IoT) sensing. Background Technology

[0002] In semiconductor manufacturing, especially in cleanrooms for wafer fabrication at 3nm and below nodes, the control requirements for gaseous molecular contaminants are extremely stringent. During operation or start-up and shutdown, production equipment instantaneously releases trace amounts of AMCs, such as ammonia and acid gases, accompanied by powerful and irregular dynamic thermal plumes. These thermal plumes carry AMCs and diffuse rapidly; once they come into contact with the wafer surface, they can cause the entire batch of products to be scrapped.

[0003] Existing technologies primarily employ a full-space dilution purification strategy, which dilutes pollutant concentrations by increasing the air exchange rate throughout the cleanroom. The inherent drawback of this method is twofold: firstly, the global increase in ventilation volume to address localized, transient contamination leads to system energy consumption increasing exponentially by the cube of the air velocity (…). First, the air quality deteriorates exponentially, resulting in extremely low energy efficiency. Second, the response speed is slow, making it impossible to effectively intercept pollutants before they spread. Third, excessively high wind speeds may disrupt the laminar flow in the work area, stirring up secondary pollutants from equipment or ground sediment, creating a vicious cycle.

[0004] Against this backdrop, a sharp technical contradiction emerges: to achieve rapid capture of instantaneous AMC carried by dynamic thermal plumes—that is, to pursue extremely high local purification accuracy—ultra-high airflow velocities must be used for suppression. However, this inevitably leads to a sharp deterioration in energy consumption and an increased risk of secondary contamination. Existing technologies adhere to the design philosophy that static uniform laminar flow is the optimal solution for cleanrooms, lacking an effective solution for the problem of optimal energy removal of dynamic, heterogeneous, and instantaneous contaminants. Therefore, a new control system that can overcome this technical contradiction and achieve synergistic optimization of energy consumption and purification accuracy is urgently needed.

[0005] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a smart building sampling and detection control system based on Internet of Things (IoT) sensing to solve the problems mentioned in the background art.

[0007] The technical solution of the present invention includes an environmental state sensing unit, a predictive analysis unit, a flow field preprocessing unit, a path planning unit, and a cooperative control unit. The environmental condition sensing unit is used to collect environmental condition data of the target monitoring area to generate real-time data vectors; The predictive analysis unit is used to receive real-time data vectors and perform diffusion trajectory deduction according to a preset state space paradigm to generate a predicted state vector. The flow field preprocessing unit is used to perform vortex kinetic energy deconstruction processing on the thermal plume in response to the predicted state vector, so as to generate a weakened flow field environment. The path planning unit is used to combine the predicted state vector with the weakened flow field environment to perform quantitative planning analysis of the guiding path in order to generate a clearing signal. The collaborative control unit is used to generate a virtual air duct topology based on the clearing signal, and to generate flow field execution commands based on the virtual air duct topology.

[0008] Preferably, the process by which the environmental state sensing unit collects environmental state data includes: Obtain the three-dimensional motion vector of gaseous molecular pollutants; Obtain the surface temperature distribution of the equipment; To obtain the local concentration of gaseous molecular pollutants; It integrates three-dimensional motion vectors, surface temperature distribution, and local concentration of gaseous molecular pollutants to generate real-time data vectors.

[0009] Preferably, the process of generating the predicted state vector is as follows: The current system state vector, the real-time data vector generated by the environmental state sensing unit, and the control vector of the previous control cycle are input into a preset nonlinear mapping function for calculation to generate a predicted state vector. The system state vector includes the location and concentration information of gaseous molecular pollutants.

[0010] Preferably, the process of deconstructing the vortex kinetic energy is as follows: The predicted thermal plume intensity is determined based on the predicted state vector; Based on the predicted thermal plume intensity, the vibration frequency of the micro-turbulence array is adjusted to determine the active dissipation coefficient; Based on the active dissipation coefficient, the total kinetic energy decay of the large-scale vortex of the thermal plume is accelerated to generate a weakened flow field environment.

[0011] Preferably, the quantitative planning and analysis process of the guiding path is as follows: The purification ineffectiveness index is determined based on the predicted state vector; Determine the energy consumption of the drive airflow control node and normalize the energy consumption to determine the energy cost; Obtain the preset energy cost weighting factor and purification risk weighting factor; The energy cost and the purification ineffectiveness index are weighted and summed to generate a comprehensive cost function, and then a purification signal is generated based on the comprehensive cost function.

[0012] Preferably, the energy cost is a value obtained by normalizing the energy consumption of driving a single airflow control node to generate the required airflow with a preset reference energy value; The purification ineffectiveness index is the ratio of the predicted residual concentration of pollutants after passing through a single airflow control node to the preset safety threshold concentration.

[0013] Preferably, the process of generating the clearing signal includes: The comprehensive cost function is compared and analyzed with the preset cost threshold; If the overall cost function is greater than the cost threshold, an emergency cleanup signal is generated, and the cleanup risk weight factor is set to be greater than the energy cost weight factor. If the overall cost function is not greater than the cost threshold, a regular cleanup signal is generated, and the energy cost weighting factor is set to be greater than or equal to the cleanup risk weighting factor.

[0014] Preferably, the process of generating the flow field execution command is as follows: Calculate the air supply pulsation angular frequency for each unit in the vector air supply unit array; The initial phase is calculated for each unit; The flow field execution command is generated by combining the calculated air pulsation angular frequency and initial phase for each unit.

[0015] Preferably, it also includes a flow field execution unit; The flow field execution unit is used to receive and execute flow field execution commands to generate a virtual air duct to enclose and guide pollutants, thereby capturing the pollutants.

[0016] This invention provides an improved smart building sampling and detection control system based on Internet of Things (IoT) sensing, which has the following improvements and advantages compared with the prior art: 1. By fusing the three-dimensional motion vector of gaseous molecular pollutants, the surface temperature distribution of equipment, and the local concentration of gaseous molecular pollutants, a real-time data vector containing multi-dimensional physical information is generated. This high-fidelity data input provides the predictive analysis unit with a complete understanding of the pollution situation. The predictive analysis unit further inputs the real-time data vector, the current system state vector, and the control vector from the previous control cycle into a preset nonlinear mapping function. This allows for the accurate deduction of the future diffusion trend of pollutants, generating a predicted state vector, based on a full consideration of the system's own evolution and the influence of control behavior. This predictive mechanism transforms the entire control system from a passive response to an active intervention, gaining valuable decision-making lead time for subsequent precise removal, and ensuring the real-time nature of the prediction due to the use of a pre-trained model. 2. This invention abandons the traditional approach of using strong winds for suppression; the flow field preprocessing unit responds to the predicted state vector and intervenes in the early stages of thermal plume development. By adjusting the vibration frequency of the micro-turbulence vane array, it determines the active dissipation coefficient, thereby accelerating the decay of the total kinetic energy of the large-scale vortex of the thermal plume. This vortex kinetic energy deconstruction process weakens the diffusion dynamics of pollutants at the source with extremely low energy cost, generating a more controllable and weakened flow field environment. This creates the prerequisite for subsequent low-energy guidance and capture operations, demonstrating extremely high control energy efficiency. 3. This invention achieves quantitative optimization of control decisions; the path planning unit abstracts the purification task into a solvable mathematical problem; based on the predicted state vector, the purification ineffectiveness index is determined, and the energy consumption of the driving airflow control node is normalized to determine the energy cost. The two mutually constraining indicators are weighted and summed to generate a comprehensive cost function. This quantitative analysis process enables the system to abandon fuzzy decision-making based on experience and instead calculate an optimal path that achieves the best balance between purification effect and energy consumption based on the comprehensive cost function, and generate a purification signal. Attached Figure Description

[0017] The present invention will be further explained below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of the system of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0019] Example 1 Please see Figure 1 This invention provides a smart building sampling and detection control system based on Internet of Things sensing, including an environmental state sensing unit, a predictive analysis unit, a flow field preprocessing unit, a path planning unit, and a collaborative control unit; The environmental condition sensing unit is used to collect environmental condition data of the target monitoring area to generate real-time data vectors; The predictive analysis unit is used to receive real-time data vectors and perform diffusion trajectory deduction according to a preset state space paradigm to generate a predicted state vector. The flow field preprocessing unit is used to perform vortex kinetic energy deconstruction processing on the thermal plume in response to the predicted state vector, so as to generate a weakened flow field environment. The path planning unit is used to combine the predicted state vector with the weakened flow field environment to perform quantitative planning analysis of the guiding path in order to generate a clearing signal. The collaborative control unit is used to generate a virtual air duct topology based on the clearing signal, and to generate flow field execution commands based on the virtual air duct topology. A smart building sampling and detection control system based on IoT sensing aims to solve the technical contradictions of existing technologies that increase ventilation volume globally to deal with local and instantaneous pollutants, resulting in a surge in energy consumption, slow response, and increased risk of secondary pollution. This embodiment constructs a closed-loop system integrating sensing, prediction, preprocessing, planning, and control, abandoning the traditional model of combating pollution with strong winds and instead adopting a new paradigm of guiding pollution with precise airflow. This achieves rapid and accurate removal of pollution sources without significantly increasing the total energy consumption of the system. The system includes an environmental status sensing unit, a predictive analysis unit, a flow field preprocessing unit, a path planning unit, and a collaborative control unit. It is important to emphasize that the optimal performance of this system is demonstrated in a semi-enclosed environment with relatively stable boundary conditions and known main disturbance sources, such as the semiconductor cleanroom it is designed for. The prediction accuracy of the model depends on the full learning of the main physical processes in the environment, such as the thermal plume of key equipment. For sudden strong disturbance events not covered by the model, the system will rely on real-time data from the environmental state sensing unit to respond quickly. Although the advantage of the prediction lead time will be weakened at this time, the error correction capability of the closed-loop control will still exist. Unlike existing technologies that rely on a single information source for passive response, the advancement of this invention lies primarily in its forward-looking predictive capability. The environmental state sensing unit generates a real-time data vector containing multi-dimensional physical information by fusing the three-dimensional motion vector of gaseous molecular pollutants, the surface temperature distribution of the equipment, and the local concentration of gaseous molecular pollutants. This high-fidelity data input provides the predictive analysis unit with a complete understanding of the pollution situation. The predictive analysis unit further inputs the real-time data vector, the current system state vector, and the control vector from the previous control cycle into a preset nonlinear mapping function. This allows it to accurately deduce the future diffusion trend of pollutants and generate a predicted state vector, taking into full account the system's own evolution and the influence of control behavior. This predictive mechanism transforms the entire control system from a passive response to an active intervention, gaining valuable decision-making lead time for subsequent precise removal, and ensuring the real-time nature of the prediction due to the use of a pre-trained model. The purpose of the environmental state sensing unit is to capture multi-dimensional physical information related to pollutant diffusion within the target monitoring area in real time and with high fidelity, and to integrate this information into structured data. In this embodiment, the unit is a multimodal sensing network deployed in a semiconductor cleanroom to collect environmental state data and fuse it to generate a real-time data vector. The real-time data vector refers to a comprehensive dataset containing information such as the spatial distribution, movement vector, local concentration, and surrounding environmental temperature of pollutants at the same timestamp. Its function is to provide accurate and timely real-world input for subsequent predictive analysis. The predictive analysis unit aims to proactively predict the diffusion trend of pollutants in the near future based on the current environmental conditions, providing a lead time for proactive intervention. This unit receives real-time data vectors generated by the environmental state sensing unit and performs diffusion trajectory deduction according to a preset state-space paradigm to generate a predicted state vector. The predicted state vector is a mathematical description of the spatial distribution and concentration of pollutants at a future moment, serving as the target input for flow field preprocessing and path planning. This prediction process is accomplished through a nonlinear mapping function based on a long short-term memory network. The preset nonlinear mapping function refers to a neural network model whose internal weights and bias parameters have been determined through offline supervised learning of massive historical operational data from the cleanroom. This pre-trained mechanism enables the model to capture the complex temporal patterns of fluid motion, thus allowing for rapid and accurate predictions of the future without resource-intensive real-time computation upon receiving real-time input. Pre-training here refers to offline supervised learning of the Long Short-Term Memory (LSTM) network model using a large amount of historical operational data from the cleanroom before the system is officially deployed. This historical data includes sensor readings under different operating conditions, i.e., real-time data vectors. Previously executed control commands, i.e., control vectors And the actual pollutant diffusion results caused by these conditions, i.e., the true system state. and By learning these question-answer pairings, the network model The internal weights and bias parameters are adjusted and determined to simulate and represent the true physical evolution of the system; therefore, in real-time operation, the model becomes an efficient mapping function, requiring only the current input vector ( Substitute into the formula: ; in, : Predicted state vector, the model output predicting the position and concentration of gaseous molecular pollutants at the next time step; : Nonlinear mapping function, representing the model of the physical evolution law of the system. It is a neural network model whose internal weights and bias parameters have been determined by offline supervised learning of massive historical operating data of the clean room. : The system state vector at the current moment; Real-time data vector; : Control vector of the previous control cycle; System process noise vector; Subscript: represents time; By performing a quick calculation using the above formula, an accurate predicted state vector can be obtained. This avoids the need for complex real-time fluid dynamics solutions; The flow field pretreatment unit aims to weaken the energy of the thermal plume, the main carrier of pollutants, before guiding them, and to change its flow field structure, transforming it from a highly destructive macroscopic vortex into a more controllable microscopic turbulence. This unit responds to the predicted state vector generated by the predictive analysis unit, performing vortex kinetic energy deconstruction on the thermal plume that is about to form or intensify. This processing is achieved through a micro-turbulence-disrupting array driven by piezoelectric ceramics. This array adjusts its vibration frequency according to the predicted thermal plume intensity, thereby actively increasing the dissipation of the flow field, accelerating the decay of large-scale vortex kinetic energy, and ultimately generating a weakened flow field environment. The weakened flow field environment refers to a flow field whose macroscopic vortex kinetic energy is significantly reduced and whose turbulence scale is altered, serving as a foundation for subsequent virtual duct guidance with lower energy consumption and more stable control. For the thermal plume, a key carrier of pollutant diffusion, this invention abandons the traditional approach of suppressing it with strong winds. The flow field pretreatment unit responds to the predicted state vector and intervenes in the early stages of thermal plume development. By adjusting the vibration frequency of the micro-turbulence vane array, it determines the active dissipation coefficient, thereby accelerating the decay of the total kinetic energy of the large-scale vortices in the thermal plume. This vortex kinetic energy deconstruction process weakens the diffusion dynamics of pollutants at the source with extremely low energy cost, generating a more controllable and weakened flow field environment. This creates the prerequisites for subsequent low-energy guidance and capture operations, demonstrating extremely high control energy efficiency. The path planning unit aims to comprehensively consider two core indicators—energy consumption and purification effect—to calculate an optimal path for guiding and removing pollutants, i.e., the topology of the virtual air duct. This unit combines the predicted state vector with the weakened flow field environment to perform quantitative planning analysis of the guidance path. This analysis is achieved by solving for the minimum value of the comprehensive cost function, which is a weighted sum of energy cost and purification risk. The final output of the path planning is the removal signal. The removal signal refers to the information instruction containing the optimal path node sequence, which provides a definite target for the collaborative control unit to generate specific flow field execution instructions. This invention achieves quantitative optimization of control decisions; the path planning unit abstracts the purification task into a solvable mathematical problem; based on the predicted state vector, the purification ineffectiveness index is determined, and the energy consumption of the driving airflow control node is normalized to determine the energy cost. The two mutually constraining indicators are then weighted and summed to generate a comprehensive cost function. This quantitative analysis process enables the system to abandon experience-based fuzzy decision-making and instead calculate an optimal path that achieves the best balance between purification effect and energy consumption based on the comprehensive cost function, and generate a purification signal. The purpose of the collaborative control unit is to transform the abstract path instructions output by the path planning unit into executable, precisely synchronized hardware control commands in the physical world. Based on the received clearing signal, this unit generates a virtual airflow topology. Using this topology, a pre-defined inverse solver calculates precise operating parameters for each air supply and extraction unit in the flow field execution system, generating flow field execution commands. These commands are a set of specific parameters containing information such as phase, frequency, and amplitude, which drive the flow field execution units to coordinate their actions and generate the desired focused airflow. To translate the optimal path into physical reality, the collaborative control unit, based on the clearing signal, accurately calculates parameters such as the air supply pulsation angular frequency and initial phase for each unit in the vector air supply unit array, generating flow field execution commands. The flow field execution unit receives and executes these commands, forming a virtual air duct in space through the collaborative air supply of each unit to envelop and guide pollutants. This non-contact guidance method, compared to physical isolation or global ventilation, has unparalleled flexibility, response speed, and extremely low energy consumption, ultimately achieving efficient capture of pollutants. This embodiment achieves predictive, proactive, and low-energy precise removal of dynamic pollutants through the collaborative work of the aforementioned units. By predicting pollution trajectories, pre-treating the flow field environment, planning the optimal guidance path, and generating virtual air ducts based on the phased array principle, it transforms the traditional global dilution purification mode into a local precision surgical removal mode. This resolves the sharp contradiction between local purification accuracy and total system energy consumption, achieving synergistic optimization of purification effect and energy efficiency.

[0020] The process by which the environmental condition sensing unit collects environmental condition data includes: Obtain the three-dimensional motion vector of gaseous molecular pollutants; Obtain the surface temperature distribution of the equipment; To obtain the local concentration of gaseous molecular pollutants; It integrates three-dimensional motion vectors, surface temperature distribution, and local concentration of gaseous molecular contaminants to generate real-time data vectors; In this embodiment, a specific implementation of the environmental state sensing unit is provided, and the process of collecting environmental state data is further specified. The environmental state sensing unit integrates multiple sensing modalities to acquire comprehensive information during the environmental state data acquisition process. To obtain the three-dimensional motion vector of gaseous molecular pollutants, the system releases functionalized quantum dot aerosols near key equipment. These aerosols specifically adsorb target gaseous molecular pollutants, and a high-speed fluorescence camera array uses triangulation and particle image velocimetry algorithms to analyze the three-dimensional spatial position, morphology, and motion vector field of the pollution cloud in real time. To obtain the surface temperature distribution of the equipment, an infrared thermal imaging unit deployed above the monitoring area is used to monitor the temperature field distribution of dynamic heat plumes generated by the production equipment during operation or shutdown. To obtain the local concentration of gaseous molecular pollutants, a high-sensitivity chemical sensing unit array deployed in key areas is used to provide ground-based AMC concentration calibration values. The acquired three-dimensional motion vector, surface temperature distribution, and local gaseous molecular pollutant concentration are fused to generate a real-time data vector. This fusion process integrates heterogeneous data from different sensors into a unified vector structure through timestamp alignment and spatial coordinate registration, providing comprehensive and consistent input for the prediction model. Compared to a single information source, this multimodal data acquisition and fusion method provides information on the state of pollutants with extremely high fidelity and completeness; it not only knows the location, concentration, and temperature of pollutants, but more importantly, it knows their motion vectors. This richness of information greatly improves the accuracy of subsequent diffusion trajectory prediction, laying a solid data foundation for the effectiveness of the entire closed-loop control.

[0021] The process of generating the predicted state vector is as follows: The current system state vector, the real-time data vector generated by the environmental state sensing unit, and the control vector of the previous control cycle are input into a preset nonlinear mapping function for calculation to generate a predicted state vector. The system state vector includes the position and concentration information of gaseous molecular pollutants; This embodiment is a specific implementation of the predictive state vector generation process in a predictive analysis unit; The process of generating the predicted state vector utilizes the state-space paradigm in control theory, abstracting complex fluid dynamics problems into a computable predictive model; the underlying logic lies in generating the system state vector at the current moment. Real-time data vector generated by the environmental state sensing unit and the control vector of the previous control cycle. The inputs are combined and fed into a preset nonlinear mapping function. Perform calculations to generate a predicted state vector for a very short future time step. The process is described by the following formula: ; In this formula, the parameters are defined as follows: : The system state vector at the current moment, a vector containing the currently known AMC location and concentration information. Data type: vector, source: the predicted value or initial measurement value of the previous moment; Real-time data vector, containing the AMC motion vector, temperature field and concentration information collected by the sensor at the current moment; data type: vector; source: environmental state sensing unit. : Control vector of the previous control cycle, which includes applied control parameters such as wind speed and direction of each vector air supply unit. Data type: vector. Source: Historical instruction record of the cooperative control unit. : Nonlinear mapping function, representing a model of the physical evolution of the system; data type: long short-term memory network model; source: pre-trained and determined through offline supervised learning of historical cleanroom operation data. : System process noise vector, representing the uncertainty of the model and the small unmodeled perturbations. Data type: vector. Source: random variables based on statistical characteristics. : Predicted state vector, the model output predicting the location and concentration of AMC at the next time step, data type: vector, source: calculation results of this formula; By introducing the control vector from the previous cycle As input, this embodiment enables the predictive model to understand the impact of control behavior on system evolution, thereby more accurately predicting the future state of the system under active intervention; a pre-trained LSTM model is used as the mapping function. It can effectively capture the high nonlinearity and time dependence of fluid diffusion, and compared with traditional physical model solutions, it greatly improves the prediction speed and realizes rapid forward-looking inference that meets the requirements of real-time control. To ensure the robustness and security of the system, this system also includes a fault diagnosis and security assurance mechanism; this mechanism monitors input data in real time. The system assesses the effectiveness of data collection by marking sensor data as unreliable when it exceeds reasonable thresholds or remains unupdated for an extended period. In such cases, interpolation from nearby sensor data or downgrading reliance on historical data is used for prediction. Simultaneously, the flow field execution commands generated by the collaborative control unit undergo a feasibility verification module before being issued, ensuring all parameters remain within the safe operating range of the hardware actuators. Furthermore, the system incorporates a monitoring module that continuously compares the predicted status. Compared with the actual sensing state at the next moment, If the error between the measured values ​​exceeds the preset threshold continuously, the system will determine that the model is mismatched and may trigger an alarm or automatically switch to a preset safe purification mode that does not rely on a complex model, such as global low-speed ventilation, to ensure minimum safety performance.

[0022] The process of deconstructing vortex kinetic energy is as follows: The predicted thermal plume intensity is determined based on the predicted state vector; Based on the predicted thermal plume intensity, the vibration frequency of the micro-turbulence array is adjusted to determine the active dissipation coefficient; Based on the active dissipation coefficient, the total kinetic energy decay of the large-scale vortex of the thermal plume is accelerated to generate a weakened flow field environment. This embodiment is a specific implementation of the vortex kinetic energy deconstruction process in the flow field preprocessing unit; The process of deconstructing vortex kinetic energy aims to actively accelerate the energy dissipation of large-scale vortices in a thermal plume; this process is quantitatively described by a vortex kinetic energy dissipation model improved based on turbulent energy cascade theory; to achieve this process, the system relies on the predicted state vector The temperature field information contained within is used to determine the predicted thermal plume intensity, for example, quantified as the predicted buoyancy flux; based on the predicted thermal plume intensity, the vibration frequency of the micro-turbulence vane array is adjusted. A micro-turbulence plate array refers to a group of small, fast-responding mechanical structures made of piezoelectric materials, where high-frequency vibrations can introduce minute disturbances into the fluid; vibration frequency. With active dissipation coefficient There exists a pre-defined functional relationship between them. The preset functional relationship refers to the nonlinear mapping relationship stored in the system in the form of a lookup table or polynomial function after being calibrated by high-precision fluid simulation or wind tunnel experiment. This relationship ensures that dissipation effects of a specific size can be accurately generated as needed and can be accessed quickly. Based on a determined active dissipation coefficient The system accelerates the total kinetic energy of large-scale vortices in the thermal plume. The attenuation of the flow field generates a weakened flow environment; this process is described by the following equation: ; In this formula, the parameters are defined as follows: : Total kinetic energy of large-scale vortices, characterizing the energy of macroscopic vortices in thermal plumes, dimension: joule, source: estimated from flow field data; Time, unit: second; : Vortex kinetic energy generation rate, generated by the buoyancy of the heat source, dimension: watt, source: calculated based on the predicted intensity of the thermal plume; Turbulent natural dissipation rate: Characterizes the rate of natural energy loss caused by fluid viscosity, which is transferred from large-scale eddies to small-scale pulsations and ultimately converted into heat energy. Dimension: Watt. Source: Estimated based on fluid dynamics theory. Active dissipation coefficient, an coefficient introduced by the spoiler array and related to the vibration frequency. The related additional dissipation term has the following dimensions: This coefficient is related to the total kinetic energy of the vortex. product This represents the additional energy dissipation rate introduced by active control, measured in watts; source: determined based on control commands and preset functional relationships; vibration frequency. With active dissipation coefficient There exists a pre-defined functional relationship between them. This univariate functional relationship is a simplified model obtained through high-precision simulation or wind tunnel testing under specific and representative operating conditions, such as standard background flow velocity and heat source intensity. It captures the dose-effect relationship of frequency, the most important control variable, thus achieving rapid and simple control. In actual operation, the closed-loop characteristics of the system will compensate for the model errors caused by its simplification. : Vibration frequency of the micro-spoiler array; The advantage of this implementation lies in its ability to intervene and deconstruct the energy core of the thermal plume in its early stages of formation, rather than suppressing it with strong winds after it has grown and strengthened, through precise adjustment. The system can effectively weaken the powerful thermal plume with minimal energy input and energy consumption of the baffles, significantly reducing the energy required for subsequent guidance and capture steps, demonstrating extremely high control energy efficiency. The purpose of this calibration is to establish a control command for the vibration frequency of the micro-spoiler array. With physical effects, the resulting active dissipation coefficient A stable and predictable correlation exists between them; by performing this complex fluid dynamics analysis and experiment in advance, the system can store the results as a simple function. In practical control, when the system needs to generate a dissipation effect of a specific intensity to accelerate the decay of vortex kinetic energy, i.e., in the formula... A specific one is needed in the middle. The value doesn't need to be recalculated; the required vibration frequency can be directly retrieved through this functional relationship. How much; this method simplifies complex physical processes into a single fast table lookup or function call, greatly improving the system's response speed and control efficiency.

[0023] Example 2 The quantitative planning and analysis process for guiding the path is as follows: The purification ineffectiveness index is determined based on the predicted state vector; Determine the energy consumption of the drive airflow control node and normalize the energy consumption to determine the energy cost; Obtain the preset energy cost weighting factor and purification risk weighting factor; The energy cost and the purification ineffectiveness index are weighted and summed to generate a comprehensive cost function, and then a purification signal is generated based on the comprehensive cost function. Energy cost is a value obtained by normalizing the energy consumption required to drive a single airflow control node to generate the required airflow with a preset reference energy value. The purification ineffectiveness index is the ratio of the predicted residual concentration of pollutants after passing through a single airflow control node to the preset safety threshold concentration. This embodiment is a specific implementation of the path quantitative planning and analysis process in the path planning unit, and incorporates the specific definitions mentioned above; The core of the quantitative planning and analysis process for guiding paths is to find the minimum value of the comprehensive cost function defined in the discretized spatial grid; this process is based on the predicted state vector. Determine each node on the path Purification ineffectiveness index As mentioned above, the purification ineffectiveness index The pollutants pass through a single airflow control node. The ratio of the predicted residual concentration to the preset safety threshold concentration; the preset safety threshold concentration is the upper limit of AMC concentration that cannot be exceeded, set according to the stringent requirements of semiconductor manufacturing processes, such as the SEMI standard. It is a benchmark value to ensure product yield. At the same time, the system determines the drive airflow control node. Energy consumption The energy cost is then normalized to determine the energy consumption required to drive a single airflow control node, such as a vector air supply unit, to generate the required airflow. Compared with the preset reference energy value The dimensionless value obtained after normalization; the preset reference energy value. It is a standardized energy benchmark, such as the energy consumption of a single fan filter unit in a standard operating cycle. Its function is to eliminate the dimensional differences in energy consumption of different execution units, so that cost calculations are comparable. The system obtains the preset energy cost weighting factor. and purification risk weighting factor These two weights are the core adjustable parameters of the system. They are preset during system initialization based on the overall operating strategy, such as energy saving priority or performance priority, and satisfy the following conditions: Normalization constraints; The system performs a weighted summation of energy costs and purification inefficiency index to generate a comprehensive cost function. And adopting an improved A The search algorithm searches in the discretized space for paths that minimize the total cost. The minimum node sequence is used to generate a clearing signal, and then the optimal sequence is used to generate another clearing signal. The overall cost function is as follows: ; In this formula, the parameters are defined as follows: : Refers to the optimal path consisting of a sequence of nodes; : Index of discrete airflow control nodes on the path; Energy cost weighting factor, dimensionless, source: system preset; : Purification risk weighting factor, dimensionless, source: system preset; Energy consumption of node i, in joules, source: calculated based on required airflow parameters; Reference energy value, in joules, source: system preset benchmark; Purification ineffectiveness index, dimensionless, source: based on predicted state vector. calculate; This implementation unifies the conflicting objectives of purification accuracy and energy consumption within a framework through a quantified and optimizable cost function. The specific definitions described above ensure that each component of the cost function has a clear physical meaning and comparability. This enables the system to make truly optimal decisions and find a mathematically provable purification path that achieves the best balance between energy consumption and purification effect, thereby elevating the control strategy from experience-based fuzzy decision-making to data-driven quantitative optimization. Weighting factors and The adjustment of this factor is key to the system's adaptive decision-making; in the comprehensive cost function: ; and They respectively control the importance of energy costs and pollution risks in the total cost calculation; When the system determines that the risk of contamination is not high, it will generate a regular cleanup signal and set... Greater than or equal to ,For example or At this point, energy costs dominate the total cost, and the path planning algorithm will tend to find a path that, while not the fastest in terms of purification speed, is the most energy-efficient. Conversely, when the system predicts a high-risk pollution event by comparing the cost function value with a preset threshold, it will generate an emergency cleanup signal and set... Much larger ,Right now This forces the algorithm to prioritize reducing the purification ineffectiveness index during pathfinding. This means that no matter the energy cost, we must plan the fastest and most thorough removal path to ensure production safety.

[0024] The process of generating a clear signal includes: The comprehensive cost function is compared and analyzed with the preset cost threshold; If the overall cost function is greater than the cost threshold, an emergency cleanup signal is generated, and the cleanup risk weight factor is set to be much greater than the energy cost weight factor. If the overall cost function is not greater than the cost threshold, a regular cleanup signal is generated, and the energy cost weighting factor is set to be greater than or equal to the cleanup risk weighting factor. This embodiment is a further adaptive optimization of the path planning process, specifically reflected in the introduction of a dynamic adjustment mechanism in the generation of the clearing signal; In this embodiment, the process of generating the clearing signal adds a decision logic layer; to make a decision, the comprehensive cost function calculated above is used. The final value is compared and analyzed with a preset cost threshold; the preset cost threshold is a critical value determined based on risk assessment. For example, the threshold can be set to the cost value corresponding to a high-risk contamination event that is identified through historical data analysis as causing the product batch to be scrapped. The decision-making logic is as follows: If the comprehensive cost function If the value exceeds the cost threshold, it indicates that the system has predicted a high-risk pollution event. In this case, the system will generate an emergency cleanup signal and set the cleanup risk weighting factor to be much larger than the energy cost weighting factor. This forces the path planning algorithm to prioritize purification effectiveness, planning the fastest and most thorough removal path. If the overall cost function is not greater than the cost threshold, it indicates that the pollution is within a controllable range. At this time, the system generates a regular purification signal and sets the energy cost weight factor to be greater than or equal to the purification risk weight factor. or This allows the system to plan its path with energy conservation as the primary goal, while ensuring basic purification effects. This implementation endows the system with intelligent and adaptive operating mode switching capabilities; it can operate in energy-saving mode under normal conditions to minimize daily energy consumption; and when a serious pollution threat is detected, it can automatically and instantly switch to emergency mode to ensure production safety at all costs; this dynamic adjustment strategy enables the system to further optimize energy efficiency on a macro time scale, while ensuring robustness and rapid response capabilities to sudden and severe events.

[0025] Example 3 The process of generating flow field execution commands is as follows: Calculate the air supply pulsation angular frequency for each unit in the vector air supply unit array; The initial phase is calculated for each unit; The flow field execution command is generated by combining the calculated air pulsation angular frequency and initial phase for each unit. It also includes a flow field execution unit; The flow field execution unit receives and executes flow field execution commands to generate a virtual airflow channel that envelops and guides pollutants, thereby capturing the pollutants. This embodiment is a specific implementation of the process of generating flow field execution instructions by the collaborative control unit, and incorporates the aforementioned flow field execution unit for receiving and executing the instructions; The process of generating flow field execution commands is the reverse solution of airflow modulation technology; when the cooperative control unit receives the clearing signal generated by the path planning unit, i.e. the optimal node path, the internal reverse solver is activated. This solver, for each target point on the path, calculates in reverse the parameters of the multiple air sources required to form a focused airflow; this is for each unit in the vector air supply unit array. Calculate the required air supply amplitude vector. Air pulsation angular frequency and initial phase The vector air supply unit array is the core component of the flow field execution unit, consisting of multiple airflow nozzles whose wind direction, speed, and pulsation frequency can be independently controlled. Finally, combining the parameters calculated for each unit, the system generates a complete flow field execution command. The goal of this command is to cause the airflow waves emitted by each unit to undergo constructive interference at the target path point in the far-field space, forming a highly directional airflow beam. The focusing airflow formed at that point is created by the superposition of vectors as described by the following equation: ; In this formula, the parameters are defined as follows: : Composite airflow velocity vector at the target point and time The synthesis rate, dimensionless: meters per second, source: calculation results of this formula; : Index and total number of air supply units; : The air supply amplitude vector of the j-th unit, dimension: m / s, source: calculated by the inverse solver; Wave vector, related to airflow direction, source: determined based on the unit location and the geometric relationship between the target point; The pulsating angular frequency and initial phase of the air supply are the core solution outputs of this process, which are obtained from the inverse solver. The system also includes a flow field execution unit; this unit physically implements control and is used to receive and execute the aforementioned flow field execution commands; it consists of a vector air supply unit array and a distributed micro exhaust port network; upon receiving the command, each vector air supply unit works collaboratively according to the specified parameters to generate a virtual air duct to envelop and guide pollutants; this air duct is an invisible pipe formed by precisely focused airflow, which completely envelops the pollutant cloud and stably transports it along the optimal path to the synchronously activated micro exhaust ports, thereby capturing and discharging the pollutants; This reverse solver performs a reverse engineering calculation; it solves the problem of given a set of target points along the optimal node path. The core component of the focusing airflow formed at the point can be approximated by the following idealized wave control model: ; It should be noted that this formula is a simplified physical model of cooperative control, used to guide the inverse solver in calculating core parameters such as phase and frequency. It assumes that the mutual interference of airflows in the far field plays a dominant role. The actual flow field is driven by the command calculated by this model and corrected by the closed-loop feedback of the sensing unit. Each air supply unit in the above formula The required air supply pulsation angular frequency is calculated in reverse. and initial phase These two core control variables; the calculated parameters combined together constitute the final, precisely executable flow field execution instructions issued to the hardware; In this embodiment, the inverse solver is implemented based on an iterative optimization algorithm; it will solve all target points on the path. At the point, by formula The difference between the calculated synthetic airflow velocity vector and the desired guided velocity vector is calculated. The solver searches for the optimal vector within a parameter space that satisfies physical constraints, such as maximum wind speed and frequency range, using algorithms such as covariance matrix adaptive evolution strategies or particle swarm optimization. The combination process continues until the objective function converges. This process is completed in real time during control command generation. This implementation details the transformation process from abstract path to physical airflow; by calculating precise operating parameters for each execution unit, the system can sculpt airflow structures of arbitrary shapes in three-dimensional space with extremely high precision and flexibility; the existence of the flow field execution unit enables this ingenious control to be physically realized; this virtual air duct technology guides and captures pollutants in a non-contact, adaptive manner. Compared with control using physical baffles or global ventilation, it has extremely low energy consumption, extremely fast response speed, and extremely high flexibility, which is the core technology of this invention to achieve efficient, low-consumption, and precise removal; The system also possesses intelligent adaptive decision-making capabilities. By comparing and analyzing the comprehensive cost function value with a preset cost threshold, the system can autonomously determine the risk level of a pollution event. When a high-risk event is predicted, the system generates an emergency cleanup signal, setting the purification risk weight factor higher than the energy cost weight factor to ensure the cleanup effect at all costs. Under normal conditions, a normal cleanup signal is generated, setting the energy cost weight factor greater than or equal to the purification risk weight factor, with energy saving as the primary operational objective. This dynamic adjustment mechanism enables the system to maximize energy efficiency on a macroscopic time scale while ensuring safety redundancy. This invention constructs a complete closed-loop control system from perception and prediction to decision-making and execution by deeply coupling environmental state sensing, predictive analysis, flow field preprocessing, path planning and collaborative control. It replaces the high-energy-consuming global dilution with predictive, low-disturbance precise airflow guidance, transforms the control mode of pollutants, and synergistically optimizes the purification effect and energy efficiency, which has significant technological advancements. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A smart building sampling and detection control system based on IoT sensing, characterized in that, It includes an environmental state sensing unit, a predictive analysis unit, a flow field preprocessing unit, a path planning unit, and a collaborative control unit; The environmental condition sensing unit is used to collect environmental condition data of the target monitoring area to generate real-time data vectors; The predictive analysis unit is used to receive real-time data vectors and perform diffusion trajectory deduction according to a preset state space paradigm to generate a predicted state vector. The flow field preprocessing unit is used to perform vortex kinetic energy deconstruction processing on the thermal plume in response to the predicted state vector, so as to generate a weakened flow field environment. The path planning unit is used to combine the predicted state vector with the weakened flow field environment to perform quantitative planning analysis of the guiding path in order to generate a clearing signal. The collaborative control unit is used to generate a virtual air duct topology based on the clearing signal, and to generate flow field execution commands based on the virtual air duct topology.

2. The smart building sampling and detection control system based on IoT sensing according to claim 1, characterized in that, The process by which the environmental state sensing unit collects environmental state data includes: Obtain the three-dimensional motion vector of gaseous molecular pollutants; Obtain the surface temperature distribution of the equipment; To obtain the local concentration of gaseous molecular pollutants; It integrates three-dimensional motion vectors, surface temperature distribution, and local concentration of gaseous molecular pollutants to generate real-time data vectors.

3. The intelligent building sampling and detection control system based on IoT sensing according to claim 1, characterized in that, The process of generating the predicted state vector is as follows: The current system state vector, the real-time data vector generated by the environmental state sensing unit, and the control vector of the previous control cycle are input into a preset nonlinear mapping function for calculation to generate a predicted state vector. The system state vector includes the location and concentration information of gaseous molecular pollutants.

4. The intelligent building sampling and detection control system based on IoT sensing according to claim 1, characterized in that, The process of deconstructing the vortex kinetic energy is as follows: The predicted thermal plume intensity is determined based on the predicted state vector; Based on the predicted thermal plume intensity, the vibration frequency of the micro-turbulence array is adjusted to determine the active dissipation coefficient; Based on the active dissipation coefficient, the total kinetic energy decay of the large-scale vortex of the thermal plume is accelerated to generate a weakened flow field environment.

5. A smart building sampling and detection control system based on IoT sensing according to claim 1, characterized in that, The quantitative planning and analysis process for the guidance path is as follows: The purification ineffectiveness index is determined based on the predicted state vector; Determine the energy consumption of the drive airflow control node and normalize the energy consumption to determine the energy cost; Obtain the preset energy cost weighting factor and purification risk weighting factor; The energy cost and the purification ineffectiveness index are weighted and summed to generate a comprehensive cost function, and then a purification signal is generated based on the comprehensive cost function.

6. A smart building sampling and detection control system based on IoT sensing according to claim 5, characterized in that, The energy cost is a value obtained by normalizing the energy consumption of driving a single airflow control node to generate the required airflow with a preset reference energy value. The purification ineffectiveness index is the ratio of the predicted residual concentration of pollutants after passing through a single airflow control node to the preset safety threshold concentration.

7. A smart building sampling and detection control system based on IoT sensing according to claim 5, characterized in that, The process of generating the clearing signal includes: The comprehensive cost function is compared and analyzed with the preset cost threshold; If the overall cost function is greater than the cost threshold, an emergency cleanup signal is generated, and the cleanup risk weight factor is set to be greater than the energy cost weight factor. If the overall cost function is not greater than the cost threshold, a regular cleanup signal is generated, and the energy cost weighting factor is set to be greater than or equal to the cleanup risk weighting factor.

8. A smart building sampling and detection control system based on IoT sensing according to claim 1, characterized in that, The process of generating the flow field execution command is as follows: Calculate the air supply pulsation angular frequency for each unit in the vector air supply unit array; The initial phase is calculated for each unit; The flow field execution command is generated by combining the calculated air pulsation angular frequency and initial phase for each unit.

9. A smart building sampling and detection control system based on IoT sensing according to claim 1, characterized in that, It also includes a flow field execution unit; The flow field execution unit is used to receive and execute flow field execution commands to generate a virtual air duct to enclose and guide pollutants, thereby capturing the pollutants.

Images