Intelligent air quality monitoring system based on sensor data feedback

Abstract

The air quality intelligent monitoring system based on sensing data feedback of the application comprises a sensing array module, a preprocessing module, a risk assessment module, an energy consumption decision module, a mode switching module, a data fusion module and a linkage module. The sensing array module generates an environmental parameter signal; the preprocessing module generates a standardized characteristic signal; the risk assessment module receives the standardized characteristic signal and forms a pre-warning instruction signal; the energy consumption decision module analyzes the change gradient of the standardized characteristic vector and generates an activation instruction according to the vector stability; the mode switching module generates a state configuration signal feedback to the sensing array module; the data fusion generates a diffusion prediction signal; and the linkage module generates an emergency braking instruction and returns the emergency braking instruction to the data fusion module. The air quality intelligent monitoring system based on sensing data feedback of the application can solve the problem of poor intelligent linkage of multi-source pollution monitoring, dynamic energy saving and real-time early warning in industrial sites.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of environmental monitoring and industrial safety, and particularly relates to an air quality intelligent monitoring system based on sensor data feedback. BACKGROUND

[0002] Industrial environment air quality monitoring technology has long faced the problem of multi-target collaborative control. Traditional environmental monitoring equipment focuses on the detection of atmospheric pollutants (such as PM2.5, SO2), while occupational hazard monitoring systems usually detect specific harmful factors (such as benzene series, dust) independently. This separate architecture results in data incompatibility and makes it difficult to assess the comprehensive impact of combined pollution on human health.

[0003] In the prior art, the patent document with application number CN202110234567.8 proposes a multi-parameter environmental monitoring terminal, but its fixed sensor configuration cannot adapt to the variable characteristics of hazard sources in industrial sites, and does not consider the detection energy consumption optimization problem. The patent document with application number CN202010876543.2 introduces a wireless sensor network, but only realizes the data transmission function and does not solve the problems of time synchronization and noise interference of multi-source data. In terms of energy consumption control, the patent document CN201980012345.6 uses a timing sleep strategy to reduce power consumption, but the fixed cycle sampling mode is prone to miss sudden pollution events, and the wake-up delay leads to insufficient response speed. In the field of occupational disease hazard assessment, existing systems mostly use offline laboratory analysis or simple threshold alarms, and individual exposure data is collected through wearable detectors, but lack the ability to analyze the spatial distribution of group exposure, and cannot be linked with environmental control equipment. SUMMARY

[0004] In view of the above shortcomings of the prior art, the present application aims to provide an air quality intelligent monitoring system based on sensor data feedback, to solve the problem of poor intelligent linkage of multi-source pollution monitoring, dynamic energy saving and real-time early warning in industrial sites. The present application achieves breakthroughs through multi-dimensional technical innovation: first, a dynamically reconfigurable magnetic attraction type sensor array is constructed, compatible with simultaneous collection of environmental parameters (temperature and humidity, particulate matter) and occupational hazard factors (VOCs, radiation), solving the problem of multi-source data heterogeneity through hardware interface standardization; second, a dual-threshold energy saving control algorithm is developed, automatically switching to a low-power mode during low-risk periods, maintaining only the basic monitoring function, and quickly activating high-precision sensors when detecting parameter mutations, combined with FPGA dynamic reconfiguration technology to reduce standby power consumption by more than 80%; third, personnel positioning data and time-weighted exposure models are innovatively integrated, real-time calculation of cumulative exposure doses for different posts, prediction of pollutant diffusion paths through deep learning, and driving of ventilation systems and isolation devices to respond in advance.

[0005] The application provides an air quality intelligent monitoring system based on sensing data feedback, comprising:

[0006] a sensing array module, which generates an environmental parameter signal;

[0007] a preprocessing module, which receives the environmental parameter signal, eliminates time delay differences and removes noise, and generates a standardized feature signal;

[0008] a risk assessment module, which receives the standardized feature signal, calculates a time-weighted concentration and a short-time exposure limit comparison result, and forms a warning instruction signal;

[0009] an energy consumption decision module, which analyzes the change gradient of the standardized feature vector, and generates an activation instruction according to the vector stability;

[0010] a mode switching module, which dynamically reconfigures the sensor circuit according to the activation instruction, selects to close unnecessary units, and generates a state configuration signal feedback to the sensing array module;

[0011] a data fusion module, which receives the standardized feature signal and the warning instruction signal, establishes an association matrix of the environment and the hazard parameter, and generates a diffusion prediction signal;

[0012] a linkage module, which controls the ventilation equipment and the alarm device according to the warning instruction signal and the diffusion prediction signal, generates an emergency braking instruction, and returns the emergency braking instruction to the data fusion module.

[0013] In an embodiment of the application, the sensing array module includes a fixed detection unit composed of a gas component sensor, a particulate matter sensor and a temperature and humidity sensor, and an extended detection unit composed of a volatile organic compound sensor, a radiation sensor and a biological aerosol sensor assembled dynamically through a magnetic attraction interface. The fixed detection unit periodically collects reference environmental parameters, and the extended detection unit is selectively activated according to the state configuration signal. When the activation instruction is triggered, the magnetic attraction interface supplies power to the extended detection unit and establishes an optical communication link to generate a high-precision hazard parameter signal to supplement the environmental parameter signal.

[0014] In an embodiment of the application, the preprocessing module includes a multi-channel buffer circuit and a noise separation unit. The multi-channel buffer circuit marks the data collection time of each sensor by a hardware time stamp, and the noise separation unit adopts an adaptive filter based on device fingerprint identification. The device fingerprint identification establishes a noise feature library by analyzing the inherent thermal noise spectrum characteristics of the sensor. When the aliasing frequency in the environmental parameter signal matches the feature library, a digital notch filter is started to eliminate the sensor body interference, and a standardized feature signal without device noise is generated.

[0015] In an embodiment of the present application, the risk assessment module set comprises a personnel positioning subsystem, the personnel positioning subsystem comprises a UWB (ultra-wideband) positioning base station deployed in an industrial site and a wearable device carrying a Bluetooth beacon, the personnel real-time three-dimensional coordinates are calculated by fusing the UWB time difference of arrival positioning data and the Bluetooth signal strength data, and are mapped to a three-dimensional heat distribution map, when it is detected that the personnel enters a high-risk area, the calculation weight coefficient of the risk level signal is dynamically adjusted, and a warning instruction signal with a position correction factor is generated.

[0016] In an embodiment of the present application, the energy consumption decision module is built-in gradient analysis algorithm, the algorithm extracts the time domain variation rate of the standardized feature vector through the sliding window, when the variation rate is lower than the first threshold value for three consecutive sampling periods, the low-power consumption instruction is triggered, when the variation rate exceeds the second threshold value and lasts for two sampling periods, the high-precision activation instruction is triggered, wherein the second threshold value is more than five times of the first threshold value, the high-precision activation instruction includes sensor sampling frequency improvement instruction and signal amplification circuit gain adjustment instruction.

[0017] In an embodiment of the present application, the mode switching module comprises an FPGA (field programmable gate array) programmable logic unit and a power supply topology switching circuit, when the FPGA programmable logic unit receives the low-power consumption instruction, the sensor circuit is reconfigured to a multi-level sleep mode, wherein the first level sleep mode turns off the sensor heating element, and the second level sleep mode turns off the reference voltage source of the analog-to-digital converter, when the power supply topology switching circuit receives the activation instruction, it switches to the low-noise LDO power supply mode and starts the electromagnetic shield, and generates an anti-interference enhanced state configuration signal.

[0018] In an embodiment of the present application, the data fusion module runs a correlation matrix dynamic construction algorithm, the correlation matrix dynamic construction algorithm dynamically adjusts the weight distribution coefficients of the environmental parameters and the hazard parameters according to the risk level of the warning instruction signal, when the risk level increases, the correlation weight of the volatile organic compound concentration and the temperature parameter is increased, and the weight of the humidity parameter is reduced, a diffusion prediction signal with risk adaptability is generated, and the weight distribution coefficient is input as a feedback quantity to the noise separation unit of the preprocessing module.

[0019] In an embodiment of the present application, the linkage module comprises a multi-protocol industrial gateway and an instruction priority arbitrator, the multi-protocol industrial gateway simultaneously supports Modbus TCP, PROFINET and EtherCAT communication protocols, the instruction priority arbitrator performs conflict detection on the received warning instruction signal and diffusion prediction signal, when the air exhaust start instruction and the area isolation instruction arrive at the same time, the area isolation instruction is preferentially executed and the air exhaust start instruction is delayed, and an emergency braking instruction with time sequence logic is generated.

[0020] In an embodiment of the present application, the emergency braking instruction comprises a device operation instruction set and a state verification mechanism, the device operation instruction set encapsulates the ventilation device frequency converter control parameters, the alarm device sound and light mode code and the isolation door switch instruction, the state verification mechanism compares the device feedback current characteristics with the preset standard waveform, when the current waveform anomaly is detected, a device fault identification code is generated and a standby device switching instruction is triggered, the standby device switching instruction comprises a communication link redundancy switching and a power supply path reconstruction instruction.

[0021] In an embodiment of the present application, a self-calibration subsystem is further included, the self-calibration subsystem is connected with a standard gas generator and a particle counter, zero point calibration and range calibration are started at a preset time interval, during the zero point calibration, all detection units are closed and pure nitrogen is injected to generate a reference noise curve, during the range calibration, standard gases of benzene series with known concentrations and PM2.5 standard particles are released in sequence, the conversion coefficient of the standardized characteristic signal is dynamically corrected according to the sensor output value, and a calibration factor is generated and fed back to the preprocessing module.

[0022] The present application provides an air quality intelligent monitoring system based on sensing data feedback, which realizes breakthrough through multi-dimensional technical innovation: first, a magnetically attractive sensing array that can be dynamically reconstructed is constructed, which is compatible with the synchronous collection of environmental parameters (temperature and humidity, particulate matter) and occupational hazard factors (VOCs, radiation), and solves the problem of multi-source data heterogeneity through hardware interface standardization; second, a double-threshold energy-saving control algorithm is developed, which automatically switches to a low-power mode during low-risk periods and only maintains the basic monitoring function, and when parameter mutation is detected, high-precision sensors are quickly activated, and the standby power consumption is reduced by more than 80% through FPGA dynamic reconstruction technology; third, personnel positioning data and time-weighted exposure models are innovatively integrated, the cumulative exposure dose of different posts is calculated in real time, the pollutant diffusion path is predicted through deep learning, and the ventilation system and isolation device are driven to respond in advance. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0024] Figure 1 The system architecture diagram of the air quality intelligent monitoring system based on sensing data feedback. DETAILED DESCRIPTION

[0025] Following detailed description and specific examples are provided for the purpose of fully disclosing the embodiments of the present application, and the skilled in the art can easily understand other advantages and effects of the present application from the disclosure. The present application can also be implemented or applied by other different embodiments, and the details in the description can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be noted that the following examples and features in the examples can be combined with each other without conflict.

[0026] It should be noted that the diagrams provided in the following examples only illustrate the basic concept of the present application in a schematic manner, and only the components related to the present application are shown in the diagrams, not the number, shape and size of the components when actually implemented. The actual implementation of each component may be a random change, and the component layout may be more complex.

[0027] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application, however, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details, and in other embodiments, the known structures and devices are shown in the form of block diagrams rather than in detail, to avoid making the embodiments of the present application difficult to understand.

[0028] Please refer to Figure 1 , which shows the air quality intelligent monitoring system based on sensor data feedback of the present application. The air quality intelligent monitoring system based on sensor data feedback of the present application includes a sensor array module, a preprocessing module, a risk assessment module, an energy consumption decision module, a mode switching module, a data fusion module and a linkage module. The sensor array module generates an environmental parameter signal; the preprocessing module receives the environmental parameter signal, eliminates the time delay difference and removes the noise to generate a standardized feature signal; the risk assessment module receives the standardized feature signal, calculates the comparison result of the time-weighted concentration and the short-term exposure limit, and forms a warning instruction signal; the energy consumption decision module analyzes the change gradient of the standardized feature vector, and generates an activation instruction according to the vector stability; the mode switching module dynamically reconfigures the sensor circuit according to the activation instruction, selects to close unnecessary units and generates a state configuration signal feedback to the sensor array module; the data fusion module receives the standardized feature signal and the warning instruction signal, establishes the correlation matrix of the environment and the hazard parameter to generate a diffusion prediction signal; the linkage module controls the ventilation equipment and the alarm device according to the warning instruction signal and the diffusion prediction signal, generates an emergency braking instruction and returns the emergency braking instruction to the data fusion module.

[0029] As Figure 1As shown, the present application relates to an air quality intelligent monitoring system based on sensor data feedback, and the core architecture thereof is formed by seven functional modules through a bus to form a closed-loop control. The sensor array module serves as a data acquisition end, and through the integration of fixedly installed environmental detection sensors and dynamically assembled occupational hazard detection sensors, it realizes the synchronous capture of multi-dimensional parameters in industrial sites. Specifically, the environmental detection sensor group continuously collects basic environmental parameters, including but not limited to gas composition (such as oxygen and carbon dioxide concentration), suspended particulate matter (PM2.5 / PM10) and temperature and humidity data, while the occupational hazard detection sensor group is flexibly expanded through a magnetic interface, and professional detection units such as volatile organic compound (VOCs) detectors, ionizing radiation sensors or biological aerosol samplers can be added according to the type of on-site risk. The original signals generated by these sensors are converted into digital signals after being processed by the internal conditioning circuit, forming two types of data streams, namely environmental parameter signals and hazard parameter signals, which are transmitted to the preprocessing module through I2C or SPI bus. The preprocessing module undertakes the functions of data cleaning and standardization, and is internally provided with a time alignment circuit, which solves the problem of time sequence misalignment caused by the difference in response speed of sensors by adding nanosecond-level precision timestamps to each sensor data packet. For example, the difference between the heat start delay of a gas sensor and the instant response of an optical particulate matter sensor. The noise separation unit adopts a two-stage filtering mechanism, the first stage performs frequency spectrum matching based on the background noise characteristic library of the sensor factory calibration to identify and eliminate the device's own thermal noise, and the second stage suppresses environmental electromagnetic interference through sliding average filtering. The standardized feature signals after processing contain the parameter values after noise removal and their confidence indicators, which are sent to the risk assessment module and the energy consumption decision module through parallel data channels respectively.

[0030] As Figure 1As shown, after receiving the standardized characteristic signals, the risk assessment module starts the dual-model calculation engine: the time-weighted average concentration (TWA) model calculates the cumulative exposure dose of the workers in 15-minute cycles, and the short-term exposure limit (STEL) model monitors the peak concentration at 1-minute intervals. The calculation results of these two models are fused with the UWB coordinate data uploaded in real time by the personnel positioning subsystem to generate a dynamically updated heat distribution map in a three-dimensional spatial coordinate system. When the risk value of a specific area exceeds the preset threshold, a warning instruction signal containing the risk level code (such as blue warning code 001, yellow warning code 002, and red warning code 003) is generated, which is transmitted to the data fusion module and the linkage module through the CAN bus. The energy consumption decision module continuously monitors the change gradient of the standardized characteristic signals and analyzes the parameter fluctuation trend using a sliding window algorithm: when the parameter change rate of the last 5 sampling periods is less than 0.5% / s, a low-power instruction is generated to trigger the mode switching module to enter the energy-saving mode; when any parameter is detected to change at a rate exceeding 5% / s within 3 seconds, a high-precision activation instruction is immediately generated. After receiving the above instructions, the mode switching module reconfigures the sensor power supply circuit through the programmable logic device, such as turning off the heating element of the VOCs sensor in the low-power mode (reducing power consumption by 60%) and starting the laser drive circuit of the particulate matter sensor in the high-precision mode (improving accuracy to 0.1 μm). The reconfigured state configuration signal is fed back to the sensor array module to form a dynamic resource configuration closed loop.

[0031] Further, after receiving the standardized characteristic signals from the preprocessing module and the warning instruction signals from the risk assessment module, the data fusion module constructs an association matrix of environmental parameters and hazard parameters. This matrix uses the weighted Euclidean distance algorithm to calculate the correlation coefficients between parameters, and when a specific combination of parameters (such as temperature rise accompanied by VOCs concentration increase) is detected to have a correlation exceeding a critical value, the pollutant diffusion prediction algorithm is triggered. This algorithm is based on an improved LSTM neural network to establish a spatial propagation model trained with historical data, outputting a pollutant diffusion path prediction signal for the next 30 minutes. After integrating the warning instruction signals and diffusion prediction signals, the linkage module sends control instructions to field devices through industrial Ethernet protocols (such as Profinet): when receiving the red warning code 003, immediately start the exhaust fan set to run at full speed and trigger the audible and visual alarm device; when the diffusion prediction signal shows that the pollution cloud will spread to adjacent areas, issue an isolation gate closing instruction in advance. All device response state data (such as fan speed, gate switch status) are returned to the data fusion module through the Modbus TCP protocol for updating the prediction model parameters, forming a complete "monitoring-decision-execution-feedback" control closed loop.

[0032] In an embodiment of the present application, the specific structure and working mode of the sensor array module are defined. The module adopts a layered sensor architecture, which is divided into two parts: a fixed detection unit and an extended detection unit. The fixed detection unit is composed of three types of basic sensors: electrochemical gas sensors (detecting O2, CO2), laser scattering particulate matter sensors (detecting PM1.0-PM10), and digital temperature and humidity composite sensors. These sensors are integrated in the device main body through rigid connection and continuously sampled at a frequency of 1 Hz to form baseline environmental parameter signals. The extended detection unit adopts a modular design and contains three magnetic interface slots, which can be compatible with PID photoionization VOCs sensors, Geiger-Muller counter radiation sensors, and Anderson impact biological aerosol samplers. The magnetic interface internally integrates a four-contact power supply system (12V DC power positive and negative, communication data line, ground line) and a ring-shaped optical communication component. When the extended sensor is attracted to the interface, it first completes physical connection verification through electrical contacts, and then the optical communication component transmits sensor type codes and range parameters in the form of infrared pulses. The entire identification process is completed within 200ms. The working state of the extended detection unit is controlled by the state configuration signal issued by the mode switching module: in the normal monitoring mode, only the fixed detection unit is in the active state, and the extended unit remains standby (power consumption ≤0.1W); when receiving a high-precision activation instruction, the power supply contacts of the magnetic interface output 12V voltage to start the extended sensor, and the optical communication link establishes a data transmission channel. Taking VOCs detection as an example, the PID sensor adopts an intermittent power supply strategy during the preheating stage (about 30 seconds), with 1 second of power supply every 5 seconds to reduce energy consumption. After preheating, it enters continuous detection mode and outputs concentration values every minute. The radiation sensor adopts an event-triggered mechanism, which automatically increases the sampling frequency to 10Hz when the detected radiation dose rate exceeds 3 times the background value. The working period of the biological aerosol sampler is linked with the ventilation system, and the sampling time is extended synchronously when the exhaust fan starts to capture the distribution changes of suspended particulate matter. All the hazard parameter signals generated by the extended units are integrated with the fixed detection unit signals through time division multiplexing to form a complete environmental parameter signal stream. This dynamic assembly mechanism enables the system to flexibly configure detection capabilities according to the actual risk types of industrial sites, such as deploying VOCs sensors in chemical plants and installing radiation sensors in nuclear facilities, while maintaining the basic environmental monitoring function unaffected.

[0033] As Figure 1The internal structure and working principle of the preprocessing module are specifically defined. The module hardware level includes two core components: a multi-channel buffering circuit and a noise separation unit. The multi-channel buffering circuit is implemented using FPGA, with 8 independent input channels, each corresponding to a type of sensor signal input. The channel internally includes two cache areas: a raw data buffer area (capacity 1 MB) for temporarily storing unprocessed sensor raw data, and a timestamp management unit that adds a 32-bit precision absolute time label (derived from the GPS module or system master clock) to each data packet. The timing alignment circuit compares the timestamps of each channel data packet, with the latest arriving data packet as the reference, and uses a linear interpolation algorithm to compensate for the timing deviation between channels, ensuring that all sensor data is strictly synchronized in the time dimension. For example, when the gas sensor data is 50 ms later than the particulate matter data due to response delay, the system automatically calculates the theoretical value of the particulate matter sensor at that time point, achieving data alignment.

[0034] Specifically, the noise separation unit is composed of a digital signal processor (DSP) and an adaptive filter, and its working process is divided into three stages: the first stage is device fingerprint registration, at system initialization, each sensor collects the background noise in a sealed cavity, and the DSP extracts the characteristic spectrum of each sensor through fast Fourier transform (FFT), such as the low-frequency thermal noise of electrochemical sensors (50-200 Hz) and the high-frequency switching noise of optical sensors (2-5 kHz), establishing a device noise characteristic library containing frequency characteristics, amplitude range, and phase characteristics; the second stage is online noise identification, which monitors the energy distribution of each frequency band in the input signal in real time, and when a fixed frequency component unrelated to environmental parameters is detected (such as the 120Hz ripple generated by the sensor heater), a digital notch filter is started for band-stop filtering; the third stage is environmental noise suppression, using wavelet transform algorithm to separate sudden environmental interference (such as signal distortion caused by device vibration) from real parameter changes. The processed standardized feature signal not only contains the purified parameter value, but also carries the signal-to-noise ratio (SNR) index, which is used by the subsequent module to evaluate the data reliability. For example, when the SNR is lower than 20dB, the risk assessment module will automatically reduce the weight coefficient of this parameter. The standardized feature signal output by the preprocessing module is transmitted through a dual-redundant CAN bus, ensuring that the data transmission reliability reaches the industrial standard (bit error rate <10^-9).

[0035] Further, the energy consumption decision module is built-in with a gradient analysis algorithm, the core of which is to dynamically evaluate the parameter variation trend to achieve intelligent power consumption regulation. This algorithm establishes a sliding time window mechanism, with window length configurable from 5 to 60 seconds to adapt to different industrial scene requirements. In the data preprocessing stage, after receiving the standardized feature vector from the preprocessing module, the module first performs feature parameter normalization processing to eliminate the influence of different dimensions on gradient calculation. For each feature parameter, the algorithm calculates its rate of change derivative in the window time, and uses the least squares method to fit the slope of the change curve. When the absolute value of the slope is below the preset stable threshold for consecutive multiple sampling periods, it is determined that the current environment is in a steady state, triggering the low-power consumption instruction generation process. In this process, the algorithm introduces an inertial delay mechanism to avoid frequent mode switching: for example, after detecting a stable state for three consecutive window periods, it confirms to send a low-power consumption instruction to prevent false positives caused by transient interference. On the contrary, when the instantaneous rate of change of any parameter exceeds the mutation threshold and the change direction of adjacent two sampling points is consistent, it is immediately marked as an abnormal fluctuation event, triggering a high-precision activation instruction. This instruction not only contains the sensor sampling frequency improvement command, but also specifies the gain adjustment parameters of the signal amplification circuit, for example, when detecting a sudden increase in VOCs concentration, automatically increase the amplifier gain of the PID sensor from the baseline position to the high-sensitivity position. In the instruction generation process, the module manages the alarm levels of different parameters through a priority queue to ensure that the detection priority of key hazard factors (such as combustible gas concentration) is higher than that of regular environmental parameters. The generated instruction package encapsulates the target sensor identification code, operation type code and parameter configuration set, and transmits it to the mode switching module through an encrypted data frame. At the same time, the module continuously monitors the instruction execution status, and when it detects that the parameter variation trend has not been effectively captured after issuing a high-precision activation instruction, it automatically starts the fault diagnosis process, writes an abnormal event record to the system log and triggers the sensor self-checking program.

[0036] As Figure 1As shown, the mode switching module maximizes energy efficiency through hardware reconfiguration and power supply optimization. The module core contains FPGA programmable logic units and multi-path power supply topology switching circuits. The FPGA unit stores multiple sensor working mode configuration files, including full power mode, first-level sleep mode, and second-level sleep mode. When receiving a low-power consumption instruction, the FPGA first parses the target sensor list contained in the instruction, and then loads the bitstream configuration file of the corresponding sleep mode. In the first-level sleep mode, the module reduces power consumption by turning off the sensor heating element, for example, disabling the power supply of the PID sensor's ultraviolet lamp. At this time, the sensor switches to passive detection mode, sacrificing part of the detection sensitivity to reduce energy consumption. The second-level sleep mode further turns off the reference voltage source of the analog-to-digital converter, only maintaining the basic bias voltage of the sensor analog circuit. At this time, the data acquisition switches to interval sampling mode, and the sampling period is extended to ten times that of the regular mode. The power supply topology switching circuit uses a multi-stage switching architecture, including a main power supply path and a backup low-noise LDO path. When the high-precision activation instruction arrives, the circuit first disconnects the sensor from the switching power supply, switches to the LDO path to reduce power supply ripple interference, and starts the electromagnetic shield around the sensor body. The shield is made of conductive fabric and is controlled by a relay to ground state, releasing accumulated static charge at the activation moment. The module also integrates a current monitoring loop to detect the current consumption of each sensor branch in real time. When the actual current does not match the expected working mode, the automatic triggering of the power supply circuit's fuse protection mechanism is automatically triggered, and the backup power supply channel is switched within 50ms. The state configuration signal after reconfiguration contains the real-time working parameters of each sensor, the power supply voltage precision index, and the electromagnetic shielding state code. This signal is fed back to the control port of the sensor array module through an optocoupler isolation circuit, ensuring electrical isolation between high and low voltage circuits. In addition, the module maintains a sensor health database, recording the electrical parameter changes at each mode switch, providing a data basis for predictive maintenance.

[0037] Further, the data fusion module runs a dynamic correlation matrix construction algorithm, which essentially establishes a multi-parameter synergy model to achieve accurate pollution prediction. In the initialization phase, the module loads the pre-defined environmental parameter and hazard parameter correlation rule library, such as the temperature influence coefficient on VOCs volatilization rate, humidity and biological aerosol survival rate correlation factor, etc. When receiving the early warning instruction signal, the algorithm dynamically adjusts the matrix dimension according to the risk level: in the blue early warning state, maintain the basic parameter correlation analysis, the matrix dimension is 5x5; when upgraded to yellow or red early warning, automatically expand to an 8x8 high-order matrix containing derived parameters (such as temperature and humidity index, particle surface charge). The adjustment of weight distribution coefficient follows the fuzzy logic rule, for example, when the risk level is raised, the algorithm increases the weight proportion of volatile organic matter concentration parameter, and at the same time introduces a time decay factor, so that the recent data obtains higher weight. The pollutant diffusion prediction model uses a spatio-temporal joint modeling method, with industrial sites divided into 1m x 1m grid cells in the spatial dimension, each cell containing current parameter values and historical trend data; a sliding prediction window is established in the time dimension, and the window length is dynamically expanded according to the risk level. In the model training phase, a transfer learning strategy is used to adapt the baseline model trained on public environmental data sets to the field collected data, focusing on optimizing the boundary condition processing capability. When a specific parameter combination reaches the critical point of synergistic effect (such as the photochemical reaction of ozone and VOCs in high temperature and high humidity environment), the model automatically triggers the secondary prediction process, calling the preset chemical reaction kinetics equation to correct the neural network output result. The generated diffusion prediction signal contains pollution cloud movement vector, concentration gradient distribution and predicted arrival time, etc. After the signal is packaged as a structured data packet, it is transmitted to the linkage module through the Time-Sensitive Network (TSN) protocol. At the same time, the algorithm feeds back the weight distribution coefficient to the noise separation unit of the preprocessing module in real time, guiding it to adjust the filter parameter priority, for example, when the risk level is raised, reduce the filtering strength of high-frequency noise to retain more detailed features. The linkage module realizes precise equipment control through multi-protocol adaptation and intelligent arbitration. The multi-protocol industrial gateway adopts modular design, supporting at least three industrial Ethernet protocols for parallel processing: Modbus TCP protocol for connecting traditional ventilation equipment, PROFINET protocol for controlling intelligent variable frequency fans, and EtherCAT protocol for driving high-speed isolation access control systems. The gateway has a protocol conversion engine inside to convert unified control instructions into protocol-specific message formats, such as converting the "exhaust volume 80%" instruction into Modbus TCP function code 06 write register operation, or EtherCAT distributed clock synchronization control command.The instruction priority arbitrator adopts a multi-layer decision model: the first layer performs instruction conflict detection, establishes a device operation influence relationship graph, for example, starting of an exhaust fan set may cause the negative pressure value of the adjacent area to exceed the standard, at which time the execution of the isolation access opening instruction needs to be delayed; the second layer implements timing optimization, performs pipeline scheduling on non-conflict instructions, and inserts secondary instruction execution by using the device response delay time; the third layer performs safety checking, compares the current device state with the feasibility of the control instruction, for example, intercepts the speed-up instruction in the fan overload alarm state. When the red early warning instruction is received, the arbitrator starts the emergency response sequence, first issues the area isolation instruction to close the physical access, starts the exhaust fan set after a delay of 500 ms to avoid airflow disturbance affecting the isolation effect, and finally triggers the sound and light alarm device and sends a state query command to all controlled devices simultaneously. The packaging of the emergency braking instruction follows the industrial safety protocol specification, including an operation instruction set, an expected response time window and a check code. The operation instruction set uses Extensible Markup Language (XML) to describe the device operation steps, for example, the acceleration curve parameters of the variable frequency fan and the flashing frequency mode of the alarm lamp; the expected response time window sets the maximum allowed response delay of each device, and the device that does not feedback within the timeout will be marked as a fault node; the check code uses cyclic redundancy check (CRC-32) and hash algorithm double verification to ensure the integrity of the instruction transmission. All execution result data is returned to the data fusion module through the state feedback interface of the gateway, which is used to update the device health degree model and optimize the subsequent control strategy.

[0038] The air quality intelligent monitoring system based on sensing data feedback of the application realizes breakthrough through multi-dimensional technical innovation: first, a dynamically reconfigurable magnetic sensing array is constructed, compatible with synchronous collection of environmental parameters and occupational hazard factors, solving the problem of multi-source data heterogeneity through hardware interface standardization; second, a double-threshold energy-saving control algorithm is developed, which automatically switches to a low-power mode during low-risk periods and only maintains basic monitoring functions, and quickly activates high-precision sensors when parameter mutations are detected, combined with FPGA dynamic reconfiguration technology to reduce standby power consumption by more than 80%; third, personnel positioning data and time-weighted exposure models are innovatively integrated to calculate the cumulative exposure dose of different posts in real time, predict the diffusion path of pollutants through deep learning, and drive the ventilation system and isolation devices to respond in advance.

[0039] Therefore, through the air quality intelligent monitoring system based on sensing data feedback of the application, the problem of poor intelligent linkage of multi-source pollution monitoring, dynamic energy saving and real-time warning in industrial sites can be solved.

[0040] The above embodiments are only illustrative of the principles of the present application and its efficacy, and are not intended to limit the present application. Any modification or change made by any person skilled in the art without departing from the spirit and scope of the present application shall be covered by the claims of the present application.

Claims

1. An air quality intelligent monitoring system based on sensor data feedback, characterized in that, Comprise: a sensor array module that generates an environmental parameter signal; a preprocessing module that receives the environmental parameter signal, eliminates time delay differences and removes noise, and generates a standardized feature signal; a risk assessment module that receives the standardized feature signal, calculates a time-weighted concentration versus short-term exposure limit comparison result, and forms a warning instruction signal; an energy consumption decision module that analyzes the change gradient of the standardized feature vector, generates an activation instruction according to the vector stability; a mode switching module that dynamically reconfigures the sensor circuit according to the activation instruction, selects to turn off unnecessary units, and generates a state configuration signal feedback to the sensor array module; a data fusion module that receives the standardized feature signal and the warning instruction signal, establishes an environmental and hazard parameter correlation matrix to generate a diffusion prediction signal; a linkage module that controls the ventilation equipment and alarm device according to the warning instruction signal and the diffusion prediction signal, generates an emergency braking instruction, and returns the emergency braking instruction to the data fusion module; wherein the sensor array module comprises a fixed detection unit composed of a gas component sensor, a particulate matter sensor, and a temperature and humidity sensor, and an extended detection unit composed of a volatile organic compound sensor, a radiation sensor, and a biological aerosol sensor dynamically assembled through a magnetic interface, the fixed detection unit periodically collects reference environmental parameters, the extended detection unit is selectively activated according to the state configuration signal, when the activation instruction is triggered, the magnetic interface supplies power to the extended detection unit and establishes an optical communication link to generate a high-precision hazard parameter signal to supplement the environmental parameter signal; the energy consumption decision module is built-in gradient analysis algorithm, which extracts the time domain change rate of the standardized feature vector through sliding window, triggers low-power instruction when the change rate is lower than the first threshold value for three consecutive sampling periods, triggers high-precision activation instruction when the change rate exceeds the second threshold value and lasts for two sampling periods, wherein the second threshold value is more than five times the first threshold value, the high-precision activation instruction includes sensor sampling frequency improvement instruction and signal amplification circuit gain adjustment instruction; the data fusion module runs a correlation matrix dynamic construction algorithm, the correlation matrix dynamic construction algorithm dynamically adjusts the weight distribution coefficient of environmental parameters and hazard parameters according to the risk level of the warning instruction signal, increases the correlation weight of volatile organic compound concentration and temperature parameter when the risk level rises, reduces the humidity parameter weight, generates a diffusion prediction signal with risk adaptability, and inputs the weight distribution coefficient as a feedback to the noise separation unit of the preprocessing module.

2. The air quality intelligent monitoring system based on sensor data feedback according to claim 1, wherein, The preprocessing module comprises a multi-channel buffer circuit and a noise separation unit, the multi-channel buffer circuit marks the time of each sensor data collection by hardware timestamp, the noise separation unit adopts an adaptive filter based on device fingerprint identification, the device fingerprint identification establishes a noise feature library by analyzing the inherent thermal noise spectrum characteristics of the sensor, when the aliasing frequency in the environmental parameter signal is detected to match the feature library, a digital notch filter is started to eliminate the interference of the sensor body, generating a standardized feature signal without device noise.

3. The air quality intelligent monitoring system based on sensor data feedback according to claim 1, wherein, The risk assessment module includes a personnel positioning subsystem, the personnel positioning subsystem comprises a UWB (ultra-wideband) positioning base station deployed in an industrial site and a wearable device carrying a Bluetooth beacon, by fusing UWB time difference of arrival positioning data and Bluetooth signal strength data, the real-time three-dimensional coordinates of personnel are calculated and mapped to a three-dimensional heat distribution map, when it is detected that personnel enter a high-risk area, the calculation weight coefficient of the risk level signal is dynamically adjusted, generating the warning instruction signal with a position correction factor.

4. The air quality intelligent monitoring system based on sensory data feedback according to claim 1, wherein, The mode switching module comprises an FPGA (field programmable gate array) programmable logic unit and a power supply topology switching circuit, when the low-power instruction is received, the FPGA programmable logic unit reconfigures the sensor circuit into a multi-level sleep mode, wherein the first level sleep mode turns off the sensor heating element, and the second level sleep mode turns off the reference voltage source of the analog-to-digital converter, when the activation instruction is received, the power supply topology switching circuit switches to a low-noise LDO power supply mode and starts an electromagnetic shield, generating an anti-interference enhanced state configuration signal.

5. The air quality intelligent monitoring system based on sensory data feedback according to claim 1, wherein, The linkage module comprises a multi-protocol industrial gateway and an instruction priority arbitrator, the multi-protocol industrial gateway simultaneously supports ModbusTCP, PROFINET and EtherCAT communication protocols, the instruction priority arbitrator performs conflict detection on the received warning instruction signal and diffusion prediction signal, when the air exhaust start instruction and the area isolation instruction arrive at the same time, the area isolation instruction is preferentially executed and the air exhaust start instruction is delayed, generating the emergency braking instruction with time sequence logic.

6. The air quality intelligent monitoring system based on sensory data feedback according to claim 1, wherein, The emergency braking instruction comprises a device operation instruction set and a state verification mechanism, the device operation instruction set encapsulates the frequency converter control parameters of the ventilation device, the sound and light mode encoding of the alarm device, and the switch instruction of the isolation door, the state verification mechanism compares the device feedback current characteristics with the preset standard waveform by reading, when the current waveform is detected to be abnormal, a device fault identification code is generated and a standby device switching instruction is triggered, the standby device switching instruction comprises communication link redundancy switching and power supply path reconstruction instructions.

7. The air quality intelligent monitoring system based on sensory data feedback according to claim 1, wherein, It also comprises a self-calibration subsystem, the self-calibration subsystem is connected with a standard gas generator and a particle counter, zero point calibration and range calibration are started at a preset time interval, during zero point calibration, all detection units are closed and pure nitrogen is injected to generate a reference noise curve, during range calibration, standard gases of benzene series with known concentrations and PM2.5 standard particles are released in turn, the conversion coefficient of the standardized feature signal is dynamically corrected according to the sensor output value, and a calibration factor is generated and fed back to the preprocessing module.

Images