Environmental air quality monitoring system based on Internet of Things perception

Abstract

The invention relates to the cross technical field of environment monitoring and the Internet of Things technology, and particularly discloses an environment air quality monitoring system based on Internet of Things perception. The system comprises an acoustic radar subsystem, a pollutant concentration sensing network, an atmospheric fluid mechanics calculation engine, an Internet of Things data fusion platform and a three-dimensional migration trajectory visualization terminal. The acoustic radar subsystem inverts a local wind field and turbulence structure by using urban background noise, a pollutant concentration sensing network collects multi-parameter gas data in real time, an atmospheric fluid mechanics calculation engine is coupled with a micrometeorological field to dynamically simulate a pollutant diffusion trajectory, and an Internet of Things data fusion platform realizes multi-source data space-time alignment and feature fusion. And the visual terminal presents the three-dimensional migration path and the eddy current accumulation area. According to the technical scheme, the air retention area can be accurately recognized under the windless or weak wind condition, the hardware cost is reduced, the difference between monitoring data and human body feeling is bridged, and scientific decision support is provided for urban environment governance.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of environmental monitoring and Internet of Things (IoT) technology, specifically relating to an ambient air quality monitoring system based on IoT sensing. Background Technology

[0002] With the rapid iteration of IoT sensing technology, air quality monitoring based on environmental big data has become a core supporting means for smart cities to improve ecological governance. Traditional monitoring systems, by establishing a sensing network covering a wide area, have achieved real-time collection of concentrations of major air pollutants, providing basic data references for environmental quality assessment and pollution source supervision.

[0003] For atmospheric state perception in complex microenvironments such as urban streets and canyons, the monitoring system is required to deeply integrate atmospheric physical parameters and micro-meteorological characteristics, and achieve a technological leap from isolated point source concentration collection to dynamic field perception across the entire area.

[0004] Existing air quality monitoring technologies largely rely on discretely distributed concentration sampling points, making it difficult to analyze the stagnation zones or eddies formed by pollutants in local spaces. This results in monitoring values ​​failing to accurately explain the logical discrepancies between pedestrian perception and data compliance. Traditional systems lack high spatiotemporal resolution detection methods for microscopic wind fields under stable meteorological conditions, making it difficult to capture nonlinear atmospheric turbulence structures, thus limiting the accuracy of pollutant diffusion source tracing and transport path prediction.

[0005] Existing monitoring solutions heavily rely on expensive hardware wind measurement equipment, and the algorithm models have not yet organically integrated acoustic tomography and computational fluid dynamics mechanisms, making it impossible to achieve real-time reconstruction of the three-dimensional transport trajectory of pollutants while taking into account economic efficiency. The expectation is for an ambient air quality monitoring system based on Internet of Things (IoT) sensing. Summary of the Invention

[0006] The purpose of this invention is to provide an ambient air quality monitoring system based on Internet of Things (IoT) sensing, which can solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An IoT-based ambient air quality monitoring system includes an acoustic radar subsystem, a pollutant concentration sensing network, an atmospheric hydrodynamics calculation engine, an IoT data fusion platform, and a three-dimensional transport trajectory visualization terminal. The acoustic radar subsystem is configured to deploy multiple sensing nodes in urban street canyon areas. Each node integrates a high-sensitivity microphone array. By collecting the propagation signal of urban environmental background noise in the air, it combines acoustic tomography algorithm to invert the local atmospheric turbulence structure and wind field vector field. The pollutant concentration sensing network consists of distributed multi-parameter gas sensors, used to collect concentration data of various air pollutants in the target area in real time, and synchronously transmit the concentration data to the Internet of Things data fusion platform. The atmospheric fluid dynamics calculation engine couples the computational fluid dynamics model with the micro-meteorological field data obtained from acoustic inversion to dynamically simulate and calculate the migration trajectory and retention area of ​​pollutants in three-dimensional space. The IoT data fusion platform receives multi-source heterogeneous data from the acoustic radar subsystem and the pollutant concentration sensing network, performs spatiotemporal alignment, noise filtering, and feature fusion, and generates comprehensive environmental situation information that includes the coupling relationship between pollutant concentration distribution and micro-meteorological field. The three-dimensional transport trajectory visualization terminal is connected to the Internet of Things data fusion platform and is used to present the three-dimensional transport path, eddy current accumulation area and well-ventilated area of ​​pollutants in a visual form to assist in environmental governance decision-making and public health early warning.

[0008] Preferably, the acoustic radar subsystem utilizes naturally occurring wind noise and traffic noise in the urban environment as passive sound sources, without the need for additional sound wave signal transmission. By analyzing the propagation delay, attenuation rate, and scattering pattern of background noise between different microphones, it reconstructs the wind speed vector and turbulent kinetic energy distribution of the local area.

[0009] Furthermore, the acoustic tomography algorithm is built on a Bayesian inversion framework, which introduces prior meteorological knowledge to constrain the inversion process, ensuring that it can still output a wind field structure with physical consistency even in a low signal-to-noise ratio environment.

[0010] Furthermore, the atmospheric fluid dynamics calculation engine has a built-in lightweight computational fluid dynamics solver, which can dynamically solve the simplified form of the Navier-Stokes equations based on the real-time updated wind field vector field and boundary conditions, and deduce the diffusion, sedimentation and resuspension behavior of pollutant particles in complex street valley terrain.

[0011] Preferably, the sensor nodes in the pollutant concentration sensing network are co-located with the sensing nodes of the acoustic radar subsystem to ensure that the pollutant concentration data and the micro-meteorological field data are strictly aligned in spatial coordinates, eliminating field perception distortion caused by positional offset.

[0012] Furthermore, the IoT data fusion platform adopts an edge-cloud collaborative architecture, where preliminary data cleaning and feature extraction are completed at the edge computing unit close to the sensing node, and only key environmental features are uploaded to the cloud for global fusion and long-term trend modeling, reducing communication bandwidth requirements and improving system response speed.

[0013] Furthermore, the three-dimensional transport trajectory visualization terminal supports retrospective analysis of pollutant diffusion processes by time slices and can mark "air dead loop zones"—areas where wind field vortices persist and pollutant concentrations exceed preset thresholds, providing precise targets for urban ventilation corridor planning and pollution source control.

[0014] Preferably, even under windless or weak wind conditions, the system can still identify the local circulation structure induced by the building geometry through the acoustic radar subsystem, breaking through the limitations of traditional wind measurement relying on ultrasonic anemometers, and realizing the analysis of pollutant retention mechanisms under calm and stable weather conditions.

[0015] Furthermore, the high-sensitivity microphone array adopts an omnidirectional sound pickup design and has wide frequency response characteristics, which can capture the full spectrum of environmental sound signals from low-frequency wind vibration to mid-to-high frequency traffic noise, providing sufficient information dimensions for acoustic tomography.

[0016] Furthermore, a feedback loop is established between the atmospheric hydrodynamics calculation engine and the IoT data fusion platform. When the visualization terminal identifies an abnormal stagnation area, it can trigger local nodes to increase the sampling frequency or start directional sound source excitation to enhance the perception resolution of the abnormal stagnation area.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The ambient air quality monitoring system based on IoT sensing provided by this invention utilizes background noise to invert the microscopic wind field structure, eliminating the need for expensive dedicated wind measurement equipment and reducing the hardware cost of high-resolution meteorological sensing. By calculating the three-dimensional transport trajectory of pollutants in real time, the system can accurately identify eddy current retention areas in street canyons, scientifically explaining the phenomenon of "monitoring data meeting standards but human perception of turbidity," thus bridging the cognitive gap between objective data and subjective experience.

[0018] 2. Even under windless or weak wind conditions, the system can still analyze local circulation caused by urban building layout, improving the accuracy of pollutant diffusion source tracing and retention prediction. The system adopts an edge-cloud collaborative data processing architecture, which optimizes resource utilization efficiency while ensuring perception accuracy, providing a new monitoring paradigm for smart city environmental governance that combines economy, intelligence, and scalability. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall technical solution architecture according to the present invention; Figure 2 This is a schematic diagram of the core principle framework for local micro wind field inversion based on acoustic tomography algorithm according to the present invention. Figure 3This is a flowchart illustrating the logical process framework for spatiotemporal alignment and feature fusion of multi-source heterogeneous data using the IoT data fusion platform according to the present invention. Figure 4 This is a schematic diagram illustrating the multi-level interaction relationship and data flow between sensing nodes, edge computing units, and the cloud platform according to the present invention. Figure 5 This is a flowchart illustrating the logical process of extrapolating the three-dimensional transport trajectory of pollutants using an atmospheric hydrodynamics calculation engine coupled with microscopic meteorological field data, according to the present invention. Detailed Implementation

[0020] Example 1: Please refer to the appendix Figure 1 To be continued Figure 5 To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0021] An ambient air quality monitoring system based on Internet of Things (IoT) sensing includes an acoustic radar subsystem, a pollutant concentration sensing network, an atmospheric fluid dynamics calculation engine, an IoT data fusion platform, and a three-dimensional transport trajectory visualization terminal. The acoustic radar subsystem is configured with multiple sensing nodes for deployment in urban street canyon areas. Each sensing node integrates a high-sensitivity microphone array, an acoustic signal preprocessing unit, and a local edge inference module. The acoustic radar subsystem collects the propagation signal of urban environmental background noise in the air through the high-sensitivity microphone array. Using the acoustic tomography algorithm built into the local edge inference module, combined with the propagation characteristics of sound waves in non-uniform media, it infers the local atmospheric turbulence structure and wind field vector field. The high-sensitivity microphone array is configured as an omnidirectional sound pickup structure, with broadband response characteristics covering low-frequency wind vibration to mid-to-high frequency traffic noise, and can capture fine acoustic features in the urban background sound field.

[0022] The pollutant concentration sensing network consists of distributed multi-parameter gas sensors, signal conditioning circuits, and wireless transmission modules, which collect real-time concentration data of various air pollutants in the target area. The multi-parameter gas sensors include, but are not limited to, nitrogen dioxide sensors, sulfur dioxide sensors, carbon monoxide sensors, ozone sensors, fine particulate matter sensors, and inhalable particulate matter sensors. The pollutant concentration sensing network synchronously transmits the real-time collected concentration data to the Internet of Things data fusion platform through the wireless transmission module.

[0023] The atmospheric fluid dynamics calculation engine is physically hosted on a high-performance computing server cluster and internally runs a data-driven solver that couples computational fluid dynamics models with acoustic inversion micro-meteorological field data. The atmospheric fluid dynamics calculation engine is used to dynamically simulate and calculate the transport trajectory and retention area of ​​pollutants in three-dimensional space based on the real-time wind field vector field output by the acoustic radar subsystem. The atmospheric fluid dynamics calculation engine has the ability to handle complex street geometric boundary conditions and can analyze local circulation induced by building obstruction and heat island effect.

[0024] The IoT data fusion platform adopts an edge-cloud collaborative distributed logical architecture, consisting of a data receiving gateway, a spatiotemporal alignment processor, a feature fusion server, and a situational awareness database. The IoT data fusion platform receives multi-source heterogeneous data from the acoustic radar subsystem and the pollutant concentration sensing network. The spatiotemporal alignment processor performs coordinate system transformation and timestamp calibration on the data. The feature fusion server performs noise filtering and multi-dimensional feature extraction to generate comprehensive environmental situational information that includes the coupling relationship between pollutant concentration distribution and micro-meteorological field.

[0025] The three-dimensional migration trajectory visualization terminal is connected to the Internet of Things data fusion platform via a high-speed network. It includes a graphics rendering engine, a three-dimensional geographic information system module, and an early warning decision logic unit. The three-dimensional migration trajectory visualization terminal is used to present the three-dimensional migration path, eddy current accumulation area, and well-ventilated area of ​​pollutants in a three-dimensional graphical form. The early warning decision logic unit generates health early warning information based on preset environmental safety thresholds to assist in environmental governance decisions.

[0026] The high-sensitivity microphone array in the acoustic radar subsystem adopts a multi-aperture configuration, and each pickup unit has high signal-to-noise ratio characteristics. During operation, the acoustic radar subsystem is configured to utilize naturally occurring, incoherent passive sound sources in the urban environment for detection. These passive sound sources include, but are not limited to, broadband noise generated by wind friction against building edges, tire noise from vehicles, and powertrain noise. The acoustic signal preprocessing unit obtains the time delay information of sound wave propagation between different sensing nodes by performing cross-correlation operations. The local edge inference module calculates the equivalent sound speed distribution of the sound wave along the propagation path based on the acquired time delay information. Since the sound speed is modulated by the air flow velocity and temperature gradient, the acoustic tomography algorithm establishes a set of observation equations and uses a Bayesian inversion framework to deconvolve the observation data to separate the wind speed vector components and turbulent kinetic energy distribution parameters in the local area.

[0027] Furthermore, the acoustic tomography algorithm is configured to introduce prior meteorological knowledge constraints when performing the inversion task; the prior meteorological knowledge includes regional macro-meteorological forecast data, historical average wind field models, and geometric constraints of building layout on the flow field; the Bayesian inversion framework can output a wind field structure map with physical continuity even in the presence of environmental background noise interference by calculating the maximum value of the posterior probability distribution; the inversion process divides the target monitoring space into multiple three-dimensional voxel units, with the wind field parameters of each voxel unit as variables to be solved, and the sum of squares of the differences between the simulated sound propagation time and the measured sound propagation time is minimized through an iterative optimization algorithm.

[0028] The sensing nodes in the pollutant concentration sensing network and the sensing nodes of the acoustic radar subsystem are configured for co-location deployment, meaning they are installed on the same physical support or streetlight pole and share the same geographical coordinates. This co-location deployment ensures that the pollutant concentration sampling points and the voxel grid retrieved from the micro-meteorological field are strictly aligned in spatial coordinates. The wireless transmission module supports low-power wide-area network protocols, ensuring a reliable data link connection even in high-density urban building environments.

[0029] The computational fluid dynamics model built into the atmospheric fluid dynamics calculation engine is specifically configured as a lightweight solver based on a simplified form of the Navier-Stokes equations. The solver receives real-time wind field vectors from the acoustic radar subsystem as inlet boundary conditions and source term constraints, and discretizes the governing equations using the finite volume method or the Boltzmann method. The calculation engine can calculate the force equilibrium state of pollutant particles in complex street-valley terrain, including drag force, pressure gradient force, and gravity, and deduce the Lagrange transport trajectory of the particles. For gaseous pollutants, the calculation engine determines the evolution trend of their scalar concentration field in space by solving the convection-diffusion equations.

[0030] When processing data, the IoT data fusion platform first performs data format standardization processing through the data receiving gateway, transforming raw code streams from different manufacturers and communication protocols into a unified structured environment object. The spatiotemporal alignment processor uses high-precision timing signals to assign millisecond-level timestamps to each data point and, based on the geographic information of the sensing nodes, converts all sensed values ​​to a unified urban local coordinate system. The feature fusion server employs a multi-sensor fusion strategy to perform weighted averaging of multiple redundant observations within the same area, eliminating the drift or failure risks that may exist in a single sensor.

[0031] The graphics rendering engine of the three-dimensional transport trajectory visualization terminal supports hardware acceleration technology, enabling real-time rendering of large-scale particle flow fields and isosurface structures. The three-dimensional geographic information system module is loaded with high-precision three-dimensional models of urban buildings, which can intuitively show how pollutants flow in building gaps and how they form vortices and linger on sheltered surfaces. The early warning decision logic unit can automatically identify "air dead circulation zones," which are closed circulation areas where the wind speed is lower than a preset wind speed threshold and the pollutant concentration is continuously higher than a preset concentration threshold. Such areas are marked with a high-risk warning color on the three-dimensional interface, providing location guidance for sanitation watering operations or traffic flow control.

[0032] In a preferred configuration, the system further includes a feedback closed-loop control unit, which connects the atmospheric hydrodynamics calculation engine to the acoustic radar subsystem. When the calculation engine identifies a complex turbulent structure in a certain area that causes calculation non-convergence or increased uncertainty, the feedback closed-loop control unit sends a control command to the acoustic sensing node at the corresponding location, triggering it to increase the acoustic sampling frequency, or activating the directional sound source transmitter integrated on the node to enhance the acoustic inversion accuracy of the area by emitting controlled sound pulses of known frequency.

[0033] In terms of physical structure, each sensing node of the high-sensitivity microphone array contains at least four condenser microphone units arranged in a tetrahedral pattern, or a planar array composed of more than 16 pickup units. The array is configured to have directional sensing capability and uses beamforming technology to enhance and extract ambient noise in a specific direction. The acoustic signal preprocessing unit also includes an adaptive echo cancellation module and an environmental feature extraction submodule, used to filter out non-steady-state noise interference such as rain and thunder from complex background sounds, while retaining steady-state acoustic features related to wind fields and traffic flow.

[0034] The sensor nodes in the pollutant concentration sensing network integrate a self-calibration unit. This self-calibration unit automatically corrects the sensitivity decay caused by sensor aging by comparing the readings of adjacent sensing nodes under similar meteorological conditions. The sensor nodes also include an environmental temperature and humidity compensation unit, which uses built-in temperature and humidity sensors to correct the gas concentration measurement values ​​in real time, ensuring the accuracy of monitoring data under different seasons and climate conditions.

[0035] Example 2: To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0036] Based on Example 1, this embodiment of the ambient air quality monitoring system based on Internet of Things sensing provides an enhanced architecture based on active acoustic excitation and mobile edge computing. This enhanced architecture is particularly suitable for areas with stable meteorological conditions or highly complex terrain.

[0037] The system includes: an active enhanced acoustic sensing array, a vehicle-mounted / unmanned aerial vehicle mobile monitoring unit, a high-performance heterogeneous computing hub, a multi-dimensional situational correlation analysis platform, and an interactive early warning and response terminal.

[0038] The active enhanced acoustic sensing array, in addition to having the passive sound pickup function described in Embodiment 1, also integrates a controlled sound source generator; the controlled sound source generator is configured to emit sweeping acoustic pulses according to a preset time sequence under conditions of insufficient ambient background noise; the sensing array utilizes the principle of multipath sound wave propagation to construct a high spatiotemporal resolution acoustic attenuation spectrum by measuring the flight time difference of active sound waves between adjacent nodes, and can still detect weak local convection caused by building thermal differences even under windless conditions.

[0039] The vehicle-mounted / drone mobile monitoring unit, as a supplement to the distributed sensing nodes, is equipped with a mobile acoustic detector and a miniature chemical sensor. The mobile monitoring unit achieves dynamic positioning through the BeiDou navigation system and performs high-frequency sampling along a preset inspection path. The mobile monitoring unit transmits data back to the high-performance heterogeneous computing hub through a dedicated mobile communication link to fill the sensing gaps outside the coverage of the fixed monitoring network.

[0040] The high-performance heterogeneous computing hub adopts a parallel architecture of central processing unit and graphics processing unit, which is specifically designed to accelerate the iterative solution of computational fluid dynamics models. The computing hub contains a pollutant diffusion dynamics library, which includes sedimentation coefficients, photochemical reaction rates and binding parameters of various chemical substances with water molecules. The computing hub is configured to perform high-precision simulation of the subtle flow field in urban streets using large eddy simulation method based on real-time wind field data, and output a four-dimensional data cube containing pressure field, velocity field and concentration field.

[0041] The multidimensional situational correlation analysis platform is used to execute environmental quality assessment logic and pollution source tracing logic. The analysis platform constructs the adjoint equation of pollutant transport and uses measured concentration gradients and wind direction to reverse-calculate the possible location and emission intensity of pollution sources. The analysis platform also has a historical situational retrospective function, which can identify areas with long-term poor ventilation due to improper urban planning by replaying the three-dimensional transport process of pollutants, and provide suggested parameters for urban wind corridor optimization.

[0042] The interactive early warning response terminal is configured as a cross-platform application system, capable of providing professional-version situation views to environmental regulatory departments and mobile early warning notifications to public users. The early warning response terminal supports augmented reality display mode. When users view city streets through their smartphone cameras, the terminal can overlay virtual pollutant transport trajectories and concentration cloud maps onto the real scene, achieving an intuitive presentation of environmental information.

[0043] The active enhanced acoustic sensing array operates by executing a set of adaptive sound source excitation logic. The logic includes: real-time monitoring of the background sound pressure level; if the background sound pressure level is lower than a preset minimum detection threshold, the controlled sound source generator is activated to emit a pulse signal. The frequency range of the pulse signal is set in the ultrasonic band or a specific low-frequency band that is not sensitive to the human ear, so as to avoid noise interference to residents' lives. The pickup unit of the sensing array adopts a high dynamic range digital-to-analog converter circuit, which can simultaneously process weak active echo signals and high-energy background noise signals.

[0044] When performing a task, the vehicle-mounted / unmanned aerial vehicle (UAV) mobile monitoring unit determines its trajectory using an information gain-based path planning algorithm. The algorithm utilizes preliminary situational information fed back by fixed sensing nodes to identify sub-regions with the most drastic changes in pollutant concentration gradients or the highest uncertainty in flow field inversion, and assigns the mobile unit to these sub-regions for intensive sampling. The mobile unit and fixed nodes achieve collaborative sensing through short-range communication technology. For example, the mobile unit can act as a temporary relay node, forwarding data from fixed nodes located in blind spots.

[0045] The high-performance heterogeneous computing hub employs an adaptive mesh refinement technique when processing atmospheric fluid dynamics models. For areas with complex flow field changes, such as building edges and street intersections, the computing hub automatically increases the density of the computational mesh. For open and flat areas, a coarser mesh is used to save computational resources. The computing hub is also configured to connect in real time with the local meteorological station's numerical weather prediction system to obtain the background wind field trend for the next 6 to 24 hours, and introduce it as a large-scale boundary condition into the local micro-simulation to achieve predictive simulation of future air quality conditions.

[0046] The multi-dimensional situational correlation analysis platform integrates an expert knowledge base containing pollution emission characteristic fingerprints from different industrial sectors and transportation modes. When the system detects an abnormal increase in pollutant concentration, the analysis platform can accurately locate the pollution source by matching the measured chemical composition ratios with the fingerprints in the knowledge base and combining this with real-time inverted transport trajectories. For example, when a specific proportion of nitrogen oxides and non-methane hydrocarbons is detected and the wind field trajectory points to a traffic congestion point, the platform will automatically generate a special report on traffic pollution.

[0047] When releasing information, the interactive early warning response terminal executes different data anonymization and visualization strategies according to the recipient's role. For decision-makers, it provides a decision dashboard with detailed parameters such as accurate concentration values, wind speed vector values, and probability distribution of pollution source tracing. For ordinary citizens, it provides a sensory evaluation of air quality based on color bars and marks the "clean air corridor" near the current location in real time, which is a walking path suggestion with moderate wind speed and fresh air.

[0048] In a variant of this embodiment, the system further integrates a deep learning-based image analysis subsystem. This subsystem accesses existing security cameras in the city and uses computer vision algorithms to identify traffic flow, vehicle type distribution, and building construction status on the streets. The identified activity features are converted into pollution emission flux data and input as source terms into the atmospheric fluid dynamics calculation engine, further improving the accuracy of boundary conditions for pollutant transport simulation.

[0049] In this embodiment, the IoT data fusion platform is further enhanced into an intelligent hub with self-learning capabilities. It is internally deployed with a recurrent neural network model. By learning from massive historical monitoring data and corresponding meteorological conditions, the model can identify typical patterns of pollutant diffusion under specific meteorological combinations. When similar meteorological patterns reappear, the platform can quickly retrieve the most similar historical simulation results as initial values, significantly shortening the convergence time of the atmospheric fluid dynamics calculation engine.

[0050] Example 3: To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0051] An IoT-based ambient air quality monitoring system aims to address the challenges of redundant management of monitoring nodes and real-time field analysis under high-concurrency data streams in large-scale urban clusters. This embodiment focuses on describing the system's distributed collaboration mechanism and hardware / software co-optimization architecture.

[0052] The system includes: a massive heterogeneous sensor group, edge intelligent processing nodes, a cloud-based twin simulation hub, a cross-domain data scheduling bus, and a holographic command and display system.

[0053] The massive heterogeneous sensing group is distributed across various functional areas of the city and consists of tens of thousands of low-power acoustic radar units and chemical sensor units. Each sensing unit has an independent globally unique identifier and is automatically registered to its logical subnet through geofencing technology. Each unit in the sensing group has self-organizing networking capabilities and can maintain basic data transmission through inter-node jump transmission when the communication backbone fails.

[0054] The edge intelligent processing node is deployed in communication equipment rooms or base stations at the city block level, serving as the aggregation center for sensing groups within the region. The edge intelligent processing node is equipped with a customized inference chip, which can perform matrix operations for acoustic tomography and preliminary flow field feature extraction locally. The edge node, by executing a localized anomaly detection algorithm, only uploads environmental events that deviate from normal characteristics or compressed feature vectors to the cloud, thus alleviating the transmission pressure on the backbone network.

[0055] The cloud-based twin simulation hub is built on a high-performance cloud computing architecture. By constructing a digital twin image of the physical city, it enables parallel simulation of atmospheric motion processes at the city-wide scale. The simulation hub contains a multi-scale coupling engine that can dynamically nest the micro-meteorological field inversion results at the street level with the meso-scale meteorological background at the city level. The simulation hub is also responsible for maintaining the city-wide pollutant transport state cube, which is stored in the form of three-dimensional voxels and continuously updated with the time step.

[0056] The cross-domain data scheduling bus is responsible for efficient data routing and task load balancing between different edge nodes and the cloud hub. When a sudden pollution event occurs in a specific area, the scheduling bus can dynamically adjust computing resources and allocate more computing power to the edge nodes in the affected area to support higher frequency wind field inversion and trajectory calculation. The scheduling bus is also responsible for executing data security protocols and ensuring the transmission security of environmentally sensitive data through end-to-end encryption.

[0057] The holographic command and display system, as the top-level application interface of the system, uses immersive holographic projection technology or ultra-large screen arrays to display the urban environmental situation. The display system can not only present the visualization effect of three-dimensional movement trajectory, but also integrate a simulation exercise module. Managers can simulate and set up virtual pollution emission sources on the system, and the system will immediately predict the diffusion trend of pollutants throughout the city and the distribution of the affected population based on the current real-time meteorological field data.

[0058] When performing acoustic tomography inversion, the edge intelligent processing node employs a sparse sensing algorithm. Since urban background noise is spatially correlated, the algorithm can reconstruct the main topology of the flow field using only the time delay data of a few key acoustic paths. This processing method reduces the computational overhead of the edge node, enabling it to run on a low-power embedded processor.

[0059] The atmospheric fluid dynamics calculation module of the cloud-based twin simulation center adopts a physical constraint-based neural network method. Traditional computational fluid dynamics methods are slow when processing large-scale grids, while the physical constraint neural network achieves a fast approximate solution to physical laws by embedding the Navier-Stokes equations as part of the loss function into the deep learning model. This method improves the calculation speed of the three-dimensional transport trajectory of pollutants by more than two orders of magnitude while maintaining physical consistency, and realizes true real-time situational awareness.

[0060] The sensing units in the massive heterogeneous sensing group adopt an energy self-sufficiency design, integrating micro solar panels and high-efficiency energy storage capacitors, supporting long-term independent operation in environments without mains power supply; the outer shell of the sensing unit is made of streamlined acoustic transparent material, which can protect the internal sensitive components from wind and rain erosion, and will not cause significant obstruction or reflection interference to the propagation of sound waves.

[0061] The simulation module of the holographic command and display system further integrates population heat map data and building ventilation index; when the simulation predicts that pollutants will enter a high-density residential area or school area, the system will automatically generate the optimal evacuation suggestions or air purification facility activation plan; the system can also calculate the effect of changing traffic flow on certain road sections on improving local ventilation in the area through the interface with the traffic management system, providing a scientific basis for "controlling pollution with wind".

[0062] In a further embodiment, the system also includes an automated equipment operation and maintenance management subsystem; the equipment operation and maintenance management subsystem uses artificial intelligence technology to monitor the health status of all sensing nodes; when the acoustic characteristics or concentration readings of a certain node show persistent abnormalities and cannot be recovered through remote self-calibration, the operation and maintenance management subsystem will automatically dispatch the drone operation and maintenance unit closest to the node to inspect it, and use the camera and standard calibration source carried by the drone to perform on-site verification and automatic calibration of the node.

[0063] The cross-domain data scheduling bus employs a priority queue mechanism when processing concurrent requests; data streams involving public safety early warnings and sudden leakage accidents are given the highest priority to ensure that they can still obtain the lowest latency response when the network is congested; the bus records the lineage of all data, from the original acoustic signal to the final visualized trajectory, and the calculation logic and parameter settings of each link can be traced, ensuring the judicial impartiality and credibility of environmental monitoring data.

[0064] The feedback loop established between the atmospheric fluid dynamics calculation engine and the IoT data fusion platform is embodied in this embodiment as a dynamic perception configuration strategy: when the cloud center detects that the predicted value of pollutant concentration in a certain area deviates from the measured value by more than a preset ratio, it will automatically adjust the perception weight of the edge nodes in that area and issue new feature extraction parameters to force the edge nodes to capture more subtle pressure fluctuations or frequency drifts until the model prediction results and the measured results converge again.

[0065] This large-scale monitoring architecture based on edge-cloud collaboration, by delegating complex physical calculations to the edge and integrating macroscopic situational data to the cloud, not only achieves precise control over the microenvironment of urban streets but also possesses strong scalability, easily meeting environmental perception needs ranging from single streets to entire megacities. Through the deep integration of acoustic tomography and computational fluid dynamics models, this system constructs an intelligent neural network that "identifies pollution by listening to the wind," providing unprecedented technical means for air pollution control in modern cities.

[0066] Example 4: To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0067] This embodiment of the IoT-based ambient air quality monitoring system focuses on the application of the system in the automatic identification and intelligent diversion scheme of "air dead circulation zones" under stable weather conditions, as well as the robustness enhancement design of the acoustic radar subsystem in special acoustic environments.

[0068] The system includes: a high-precision meteorological-pollution coupled sensing array, a local circulation analysis module, an intelligent wind field guidance decision center, a multi-source heterogeneous data chain, and an adaptive terminal display unit.

[0069] The high-precision meteorological-pollution coupled sensing array is deployed in three dimensions on the balconies, rooftops and street lamp posts of street buildings, forming a multi-layered spatial sensing grid. Each node not only monitors conventional pollutants, but also integrates a pressure micro-change sensor and an infrared thermal imaging unit to assist acoustic radar in capturing the micro-thermal circulation caused by temperature differences on building facades.

[0070] The core of the local circulation analysis module lies in the microscale acoustic inversion engine, which is specifically optimized for low-speed flow fields (wind speed less than 0.5 m / s). Under stable weather conditions where conventional anemometers fail, this microscale acoustic inversion engine reconstructs weak vortex structures using a wave acoustic model by analyzing the subtle distortions of sound wave waveforms over extremely short distances. The analysis module can accurately identify the "zero-wind-speed point" in street canyons and its corresponding pollutant accumulation nuclei.

[0071] The intelligent wind field guidance decision-making center is based on a flow field intervention evaluation model, which can simulate the changing trend of pollutant transport trajectories after changing the local thermal environment or activating mechanical ventilation facilities. Based on the simulation results, the decision-making center generates the optimal air quality improvement plan and links with the city's smart infrastructure system to achieve proactive intervention in the "air dead zone".

[0072] The multi-source heterogeneous data chain employs a blockchain-based timestamp notarization technology to encrypt and notarize each frame of original audio fingerprint collected by acoustic radar. This ensures that in the event of a major pollution dispute, the wind field inversion data and pollutant trajectory projection results relied upon by the system have immutable legal validity.

[0073] The adaptive terminal display unit can automatically adjust the complexity of 3D rendering according to the display performance of the terminal device; on the mobile terminal, the display unit represents the trajectory of pollutants with a simplified vector streamline diagram; on the high-performance display wall in the command center, a high-fidelity plume diffusion model with light and shadow effects based on real-time fluid dynamics calculations is displayed.

[0074] The local circulation analysis module employs an adaptive multi-scale inversion algorithm when performing acoustic tomography. This algorithm first estimates the overall airflow direction on a coarser grid, then progressively subdivides the local grid for areas with detected rotational vector tendencies, and increases the computational density of acoustic ray tracing. In this way, the system can clearly analyze tiny eddies with diameters of only a few meters on the leeward side of buildings, which are often the root cause of long-term pollutant exceedances.

[0075] When generating intervention plans, the intelligent wind field induction decision center will comprehensively consider the ratio of energy consumption to environmental benefits. For example, when it is identified that the pollutant concentration in a street canyon is about to reach the threshold for triggering an early warning, and the weather forecast shows that the stable state will continue for more than several hours, the decision center will calculate whether turning on the outdoor units of the ventilation and air conditioning systems of specific buildings in the area or the public misting landscape can induce a sufficient pressure difference to break the existing dead circulation.

[0076] The acoustic radar unit in the high-precision weather-pollution coupled sensing array is designed to resist multipath interference in the complex reflection environment of the city. In dense building clusters, sound waves will be reflected multiple times, forming false observation paths. The sensing unit uses autocorrelation analysis and angle of arrival estimation algorithms to extract the unique direct propagation path and its first-order reflection path from the complex reflection signal, and uses these path data to enhance the geometric coverage of tomographic imaging.

[0077] The multi-source heterogeneous data link also includes an automated quality control (QC) layer, which performs a stability check on the acoustic characteristics transmitted in real time. If a non-physical sudden change in background noise is detected in a certain area, the QC layer will automatically mark the data for that period as invalid and instruct the atmospheric hydrodynamics calculation engine to perform short-term extrapolation compensation using historical stable data to ensure the continuity of the monitoring trajectory.

[0078] The system described in this embodiment fills the technological gap in the perception of micro-meteorological fields in traditional air quality monitoring by combining in-depth analysis of physical laws with modern Internet of Things sensing technology. The system can not only "see" the concentration distribution of pollutants, but also "understand" the rhythm of the wind, scientifically revealing the dynamic mechanism of pollutant diffusion, and providing solid technical support for refined urban management and healthy community planning.

[0079] In summary, this invention achieves high-precision, high-spatial-resolution monitoring of air quality in complex urban microenvironments by integrating acoustic tomography algorithms with atmospheric fluid dynamics (CFD) models and utilizing distributed IoT sensing nodes and an edge-cloud collaborative architecture. The system can analyze the transport trajectories of pollutants in three-dimensional space, identify eddy current retention zones and well-ventilated areas, thus enhancing the scientific rigor and predictability of air pollution control.

[0080] It should be understood that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Those skilled in the art can also make various modifications without departing from the scope of the present invention, and all such modifications should be included within the protection scope of the present invention.

Claims

1. An ambient air quality monitoring system based on Internet of Things (IoT) sensing, comprising an acoustic radar subsystem, a pollutant concentration sensing network, an atmospheric hydrodynamics calculation engine, an IoT data fusion platform, and a three-dimensional transport trajectory visualization terminal, characterized in that: The acoustic radar subsystem is configured with multiple sensing nodes for deployment in urban street canyon areas. Each sensing node integrates a microphone array and collects the propagation signal of urban environmental background noise in the air. It then combines the acoustic tomography algorithm to invert the local atmospheric turbulence structure and wind field vector field. The pollutant concentration sensing network consists of distributed multi-parameter gas sensors, used to collect concentration data of various air pollutants in the target area in real time, and synchronously transmit the concentration data to the Internet of Things data fusion platform. The atmospheric fluid dynamics calculation engine couples the computational fluid dynamics model with the micro-meteorological field data obtained from acoustic inversion to dynamically simulate and calculate the transport trajectory and retention area of ​​pollutants in three-dimensional space. The IoT data fusion platform is used to receive multi-source heterogeneous data from the acoustic radar subsystem and the pollutant concentration sensing network, perform spatiotemporal alignment, noise filtering and feature fusion, and generate comprehensive environmental situation information that includes the coupling relationship between pollutant concentration distribution and micro-meteorological field. The three-dimensional transport trajectory visualization terminal is connected to the Internet of Things data fusion platform and is used to present the three-dimensional transport path, eddy current accumulation area and well-ventilated area of ​​pollutants in a visual form to assist in environmental governance decision-making and public health early warning.

2. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The acoustic radar subsystem includes: A high-sensitivity microphone array, configured as an omnidirectional pickup structure, has a wideband response covering low-frequency wind vibration to mid-to-high frequency traffic noise, and is used to capture acoustic features in the urban background sound field. An acoustic signal preprocessing unit, connected to the high-sensitivity microphone array, is configured to perform cross-correlation operations to obtain the time delay information of sound wave propagation between different sensing nodes, and uses an adaptive echo cancellation module and an environmental feature extraction submodule to filter out non-steady-state noise interference from the background sound while retaining steady-state acoustic features related to wind field and traffic flow. The local edge extrapolation module incorporates the acoustic tomography algorithm, is configured to calculate the equivalent sound speed distribution of sound waves along the propagation path based on the time delay information, establish a set of observation equations, and use a Bayesian inversion framework to perform deconvolution processing on the observation data to separate the wind speed vector components and turbulent kinetic energy distribution parameters in the local area. The high-sensitivity microphone array adopts a multi-aperture distribution configuration. Each sensing node contains at least four condenser microphone units arranged in a tetrahedral pattern, or a planar array composed of multiple pickup units. It is configured to detect naturally occurring, incoherent passive sound sources in the urban environment, including broadband noise generated by wind friction with building edges, tire noise generated by vehicle movement, and power system noise.

3. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The pollutant concentration sensing network includes: Multiple distributed sensor nodes, each of which integrates a multi-parameter gas sensor, signal conditioning circuit, wireless transmission module, self-calibration unit and environmental temperature and humidity compensation unit. The multi-parameter gas sensor is used to monitor multiple components including nitrogen dioxide, sulfur dioxide, carbon monoxide, ozone, fine particulate matter, and inhalable particulate matter. The self-calibration unit is configured to automatically correct the sensitivity decay caused by sensor aging by comparing the readings of adjacent sensing nodes under similar weather conditions. The ambient temperature and humidity compensation unit uses a built-in temperature and humidity sensor to correct the measured gas concentration value in real time. Each sensing node in the pollutant concentration sensing network and the sensing node of the acoustic radar subsystem are configured to be deployed co-located, installed on the same physical support and sharing the same geographical coordinates, to ensure that the pollutant concentration sampling points and the voxel grid retrieved from the micro-meteorological field are strictly aligned in spatial coordinates.

4. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The IoT data fusion platform adopts an edge-cloud collaborative distributed logical architecture, including: The data receiving gateway is configured to perform data format standardization processing, transforming raw bitstreams from different communication protocols into a unified structured environment object; The spatiotemporal alignment processor is configured to assign millisecond-level timestamps to each data point using high-precision timing signals, and to convert all sensed values ​​to a unified urban local coordinate system based on the geographic information of the sensing nodes. The feature fusion server is configured to employ a multi-sensor fusion strategy to perform a weighted average of multiple redundant observations within the same area, eliminate measurement drift from a single sensor, and complete preliminary data cleaning and feature extraction in the edge computing unit near the sensing node. A situational awareness database is configured to store key environmental features and provide them to the cloud for global fusion and long-term trend modeling.

5. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The atmospheric hydrodynamics calculation engine is physically hosted on a high-performance computing server cluster, including: A data-driven solver is coupled with a computational fluid dynamics model, which is a lightweight solver based on a simplified form of the Navier-Stokes equations. The solver is configured to receive real-time wind field vectors from the acoustic radar subsystem as inlet boundary conditions and source term constraints, and to discretize and solve the control equations using the finite volume method or the Boltzmann method to calculate the force equilibrium state of pollutant particles in complex street valley terrain. The force equilibrium state includes drag force, pressure gradient force and gravity, and the Lagrange transport trajectory of the particles is deduced. For gaseous pollutants, the computational engine is configured to determine their scalar concentration field evolution trend in space by solving the convection-diffusion equation; The solver divides the target monitoring space into multiple three-dimensional voxel units during the inversion process. The wind field parameters of each voxel unit are used as variables to be solved. Through iterative optimization algorithms, the sum of squares of the differences between the simulated sound propagation time and the measured sound propagation time is minimized.

6. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The three-dimensional movement trajectory visualization terminal includes: The graphics rendering engine supports hardware acceleration technology and is configured to render particle flow fields and isosurface structures in real time, showing the flow of pollutants in building gaps and the vortex retention formed on the sheltered surface. The 3D geographic information system module is loaded with high-precision 3D models of urban buildings, used to achieve integrated display of geospatial data; The early warning decision logic unit is configured to automatically identify air dead circulation zones. The air dead circulation zone is defined as a closed circulation area where the wind speed is lower than a preset wind speed threshold and the pollutant concentration is continuously higher than a preset concentration threshold. The air dead circulation zone is marked with a high-risk warning color on the three-dimensional interface. The visualization terminal supports tracing the pollutant diffusion process by time slice and generates health early warning information based on preset environmental safety thresholds, assisting in urban ventilation corridor planning and pollution source control.

7. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The system also includes a feedback closed-loop control unit, which connects the atmospheric hydrodynamics calculation engine to the acoustic radar subsystem. When the atmospheric hydrodynamics calculation engine detects that the complex turbulent structure in the region increases the calculation uncertainty, the feedback closed-loop control unit sends control commands to the acoustic sensing node at the corresponding location. The control command is used to trigger the sensing node to increase the acoustic sampling frequency, or to activate the directional sound source transmitter integrated on the sensing node to enhance the acoustic inversion accuracy of the area by emitting controlled sound pulses of known frequency.

8. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The system also includes: The vehicle-mounted or drone-mounted mobile monitoring unit is equipped with a mobile acoustic detector and a miniature chemical sensor, and is configured to achieve dynamic positioning through the Beidou navigation system and to sample along a preset inspection path; The mobile monitoring unit transmits data back to the IoT data fusion platform via a mobile communication link to fill the sensing gaps outside the coverage of the fixed monitoring network. The mobile monitoring unit employs an information gain-based path planning algorithm. It utilizes the preliminary situational information fed back from fixed sensing nodes to identify sub-regions with drastic changes in pollutant concentration gradients or high uncertainty in flow field inversion, and then proceeds to those sub-regions for intensive sampling.

9. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The IoT data fusion platform is internally deployed with an intelligent hub that has self-learning capabilities. The intelligent hub includes a recurrent neural network model, configured to identify typical patterns of pollutant diffusion under meteorological combinations by learning from historical monitoring data and corresponding meteorological conditions. When a similar weather pattern reappears, the platform retrieves historical similar simulation results as initial values ​​to shorten the convergence time of the atmospheric hydrodynamics calculation engine. The computing engine also employs a physically constrained neural network method, embedding the Navier-Stokes equations as part of the loss function into the deep learning model, thereby achieving real-time approximate solutions for the three-dimensional transport trajectories of pollutants while maintaining physical consistency.

10. The ambient air quality monitoring system based on Internet of Things (IoT) sensing according to claim 1, characterized in that, The system also includes: The multidimensional situational correlation analysis platform is configured to execute pollution source tracing logic. By constructing the adjoint equation of pollutant transport, it uses measured concentration gradients and wind direction to inversely deduce the location and emission intensity of pollution sources. The multidimensional situational correlation analysis platform integrates an expert knowledge base containing pollution emission characteristic fingerprints of different industries and transportation modes. It is configured to match the measured chemical composition ratios with the characteristic fingerprints to locate the pollution source. The interactive early warning response terminal supports augmented reality display mode and can be configured to overlay virtual pollutant transport trajectories and concentration cloud maps onto real-world images. It also executes different data desensitization and visualization strategies based on the recipient's role, providing decision-makers with detailed parameter dashboards and offering the public environmental assessments and clean air corridor recommendations based on color bars.

Images