Highway tunnel ventilation and air purification intelligent control system

Abstract

The invention discloses an intelligent control system for ventilation and air purification of a highway tunnel, and relates to the technical field of tunnel environment control, and the system comprises a multi-dimensional sensing module which collects data and attaches information by means of a distributed sensor network; the multi-modal fusion module receives data, preprocesses the data, fuses the data and outputs a comprehensive result; the dual-drive coupling prediction module generates a prediction and positioning pollution source; the multi-objective optimization decision module constructs an index system selection strategy and converts an instruction; the equipment execution and feedback module drives the equipment, collects data feedback and forms a closed loop; according to the method, efficient integration analysis of tunnel multi-source data is realized through multi-dimensional perception and multi-mode fusion, the pollutant propagation prediction precision is improved through dual-drive coupling prediction, and a pollution source is accurately positioned; and through multi-objective optimization decision and closed-loop regulation, air quality, energy consumption and traffic efficiency are balanced, an intelligent scheme is provided for tunnel operation management, and system transformation is promoted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tunnel environment control, in particular to an intelligent control system for ventilation and air purification of a highway tunnel. BACKGROUND

[0002] As an important part of the transportation network, highway tunnels are key facilities for connecting different regions and relieving ground traffic pressure. With the continuous increase in traffic flow, a large amount of pollutants is generated during vehicle driving in the tunnel, which is easy to accumulate in the relatively closed space, not only reducing the air quality in the tunnel, but also affecting the visibility and posing a threat to vehicle traffic safety. At the same time, the environmental physical parameters, traffic flow state and ventilation and purification equipment operating state in the tunnel are in dynamic change, such as the alternation of peak and flat traffic flow, the fluctuation of environmental temperature and humidity, and the state attenuation during equipment operation. These dynamic factors make the pollution control in the tunnel face complex challenges. The traditional tunnel ventilation and air purification control method has been difficult to fully respond to the above complex and variable scenarios, and cannot realize real-time and accurate monitoring, effective prediction and efficient control of the pollution situation in the tunnel. Therefore, a system that can integrate multi-source information, realize intelligent analysis and collaborative regulation is urgently needed to balance the relationship between air quality protection, equipment energy saving and traffic efficiency improvement, and meet the high-quality demand of modern highway tunnel operation and management.

[0003] The traditional highway tunnel ventilation and air purification technology has obvious shortcomings in the sensing link, often only monitoring a few types of pollutants or a single dimension parameter, failing to comprehensively cover key information such as tunnel pollutant concentration, traffic flow parameter, environmental physical parameter and equipment operating state, resulting in insufficient overall understanding of the tunnel working conditions. The multi-source data collected lacks effective preprocessing mechanism and is easily disturbed by noise. At the same time, scientific spatio-temporal alignment and fusion are not carried out, making the accuracy and relevance of the monitoring data insufficient to provide reliable support for subsequent analysis. In the prediction and decision-making link, the traditional technology mainly relies on a single physical model for pollution propagation prediction, without considering dynamic variables in actual working conditions, resulting in low prediction accuracy and difficulty in locating the pollution source. The control decision often takes a single target as the guide, or only focuses on air quality while ignoring high energy consumption of equipment, or only pursues energy saving at the expense of traffic efficiency, lacking multi-target collaborative optimization thinking. In addition, the traditional technology lacks a closed-loop feedback mechanism, and the control strategy cannot be adjusted in time according to the actual effect after regulation, resulting in lagging control response and difficulty in adapting to the dynamic change of the tunnel operating environment. SUMMARY

[0004] The purpose of the present application is to make up for the deficiencies of the prior art, and provide a highway tunnel ventilation and air purification intelligent control system, which collects multiple data in the tunnel through a multi-dimensional perception module, processes through a multi-modal fusion module, predicts the spread and source of pollutants using a dual-drive coupled prediction module, based on the prediction results and real-time traffic state, a multi-objective optimization decision module constructs a three-dimensional evaluation index system, solves the optimal control strategy, finally, the equipment execution and feedback module executes the control instructions, and collects feedback data to form a closed-loop control, which effectively improves the tunnel air quality and traffic efficiency, while reducing energy consumption.

[0005] The present application provides the following technical solutions to solve the above technical problems: a highway tunnel ventilation and air purification intelligent control system, which comprises:

[0006] The multi-dimensional perception module collects the concentration of pollutants, traffic flow parameters, environmental physical parameters, abnormal sound and equipment operation state data in the tunnel through a distributed sensor network, and adds time stamps and spatial positioning information to all data;

[0007] The multi-modal fusion module receives the data output by the multi-dimensional perception module, pre-processes the data through noise reduction, time-space alignment and outlier removal, and then uses a dynamic weight fusion algorithm to fuse the multi-source data, and outputs comprehensive pollution monitoring results including pollutant concentration spatial distribution, time-space correlation characteristics and traffic flow-pollution correlation;

[0008] The dual-drive coupled prediction module generates a preliminary prediction based on the comprehensive pollution monitoring results through a flow expansion coupling algorithm, combines a multi-dimensional dynamic balance algorithm to fuse AI residual correction amounts, and outputs the pollutant spread prediction results; at the same time, a three-dimensional contribution source tracing algorithm is used to locate the pollution source;

[0009] The multi-objective optimization decision module constructs a three-dimensional evaluation index system including air quality, energy consumption and traffic efficiency based on the pollutant spread prediction results and the real-time traffic state of the tunnel, uses a dynamic weight optimization algorithm to solve the Pareto optimal control strategy set, selects a matching strategy from the optimal control strategy set based on the real-time state of the tunnel, and converts it into specific control instructions for ventilation equipment and air purification equipment;

[0010] The equipment execution and feedback module receives the cooperative control instructions, drives the operation of the axial flow fan, the jet flow fan and the electrostatic dust removal, activated carbon adsorption device; collects equipment operation parameters and air quality data after regulation and control, updates system parameters, and feeds back to the multi-dimensional perception module and the dual-drive coupled prediction module to form a closed-loop control.

[0011] Furthermore, the distributed sensor network in the multi-dimensional sensing module includes: an air parameter sensor group, with one group deployed every 50-80 meters along the longitudinal direction of the tunnel, and the two-way tunnels are deployed separately for each direction, each group containing a CO sensor, a NOx sensor, and a PM2.5 sensor; an environmental status sensor group, with infrared thermal imagers deployed at the tunnel entrance, exit, and every 1 km, and temperature and humidity sensors and visibility meters deployed every 300 meters; a traffic sensing equipment group, with a combination of high-definition cameras and millimeter-wave radar deployed 500m before the tunnel entrance, 300m after the exit, and every 800m inside the tunnel; a sound wave sensor, with one directional microphone deployed every 200 meters along the tunnel sidewall; and an equipment status sensor group, with temperature sensors installed at the motor windings of the ventilation equipment, vibration sensors installed at the bearings, and current sensors installed at the power supply of the purification equipment.

[0012] Furthermore, in the multimodal fusion module, the preprocessing steps include: applying Kalman filtering to reduce noise in air parameter sensor data, applying db4 wavelet basis 3-level decomposition to reduce noise in acoustic sensor data, and applying Gaussian filtering to reduce noise in image data; unifying non-1Hz sampling data to 1Hz through linear interpolation based on timestamps; mapping data to a 5m×5m×3m spatial grid based on tunnel mileage marker coordinates; removing outliers using the 3σ criterion; and filling missing data with a distance-weighted average of adjacent time points at the same location.

[0013] Furthermore, in the multimodal fusion module, the expression for the dynamic weight fusion algorithm is: ,in, It is a location ,time The concentration of pollutants after fusion It is spatial location. It was a moment. For the first Sensor-like sensors The monitoring value, No. Sensor-like position time Dynamic weights, It is the total number of sensor types. For multi-sensor data variance, This is the consistency correction factor.

[0014] Furthermore, in the dual-drive coupling prediction module, the expression for the flow-expanding coupling algorithm is: ,in, For position ,time Preliminary predicted concentration, for time the initial concentration of the pollutant at the location, is the sectional attenuation coefficient, is the average speed of the ventilation airflow, is the diffusion coefficient of the pollutant, is the time the pollution source intensity at the location, is the pollution source influence saturation time, is the pollution source influence coefficient, is a natural constant, is the spatial position.

[0015] Further, in the dual-drive coupled prediction module, the expression of the multi-dimensional dynamic balance algorithm is: wherein, is the final pollutant prediction concentration at the time and the position , is the prediction time step, is the pollutant type sensitivity coefficient, representing the degree of dependence of different pollutants on the physical model, is the spatial gradient adjustment factor, which strengthens the constraint of the physical model in the concentration mutation area, is the AI residual correction amount at the time and the position , is the AI correction amount confidence factor, is the preliminary prediction concentration at the position and the time .

[0016] Further, in the dual-drive coupled prediction module, the expression of the three-dimensional contribution traceability algorithm is: wherein, is the contribution traceability coefficient of the pollutant concentration of the th vehicle type at the time , the position , is the vehicle type category number, is the time window, is the integral time variable, is the traffic volume of the th vehicle type at the time , is the time decay coefficient, is the emission factor of the th vehicle type at the time , the position , is the fused pollutant concentration at the time , the position , is the first class vehicle flow, is the partial derivative of the predicted concentration of pollutants with respect to the vehicle flow of the first class vehicle flow, is the multi-modal fusion pollutant concentration at time τ at location x.

[0017] Further, the three-dimensional evaluation index system in the multi-objective optimization decision module specifically includes:

[0018] Air quality index: including pollutant concentration hard constraints and spatial distribution constraints, wherein the hard constraints are that the average CO concentration is ≤100 ppm, the NOx concentration is ≤5 ppm, the PM2.5 concentration is ≤0.05 mg / m³, and the visibility is ≥100 m every 5 minutes; the spatial distribution constraints are that the proportion of the area with pollutant concentration exceeding the threshold value in the total length of the tunnel is ≤5%, and the duration of the area with pollutant concentration exceeding the threshold value is ≤3 minutes;

[0019] Energy consumption index: including equipment operation energy consumption and start-stop energy consumption, the operation energy consumption is calculated according to the actual operation parameters of the axial flow fan, the jet flow fan, the electrostatic dust removal device, and the activated carbon adsorption device; the start-stop energy consumption is calculated according to the instantaneous energy consumption of 1.2 times the rated power of the fan each time it is started, and the instantaneous energy consumption of 0.8 times the rated power of the purification equipment each time it is switched; the total energy consumption is calculated for a period of 1 hour;

[0020] Traffic efficiency index: including vehicle flow passing efficiency and driving stability, the vehicle flow passing efficiency is that the average vehicle flow in the tunnel is ≥80% of the design flow per hour, and the average speed is ≥60 km / h; the driving stability is that the speed fluctuation coefficient is ≤0.2, the number of congestion events occurring per hour is ≤1, and the average following distance of vehicles is ≥50 m.

[0021] Further, the expression of the dynamic coupling optimization algorithm in the multi-objective optimization decision module is: wherein, is a multi-objective comprehensive optimization index, is an air quality normalization index, is an energy consumption normalization index, is a traffic efficiency normalization index, is an energy consumption coupling coefficient, is a traffic efficiency coupling coefficient.

[0022] Further, the control parameters included in the Pareto optimal control strategy set in the device execution and feedback module are:

[0023] Ventilation equipment control parameters: the number of axial flow fans to be turned on, the position to be turned on, and the speed level; the number of jet flow fans to be turned on, the directional angle, and the speed level;

[0024] Air purification equipment control parameters: number of opening sections and operating power of electrostatic dust removal device; number of opening sections, activated carbon replacement warning threshold and operating power of activated carbon adsorption device.

[0025] Compared with existing technologies, this intelligent control system for ventilation and air purification in highway tunnels has the following advantages:

[0026] I. This invention comprehensively collects various key data within the tunnel through a multi-dimensional sensing module, combined with a multi-modal fusion module's professional preprocessing and dynamic weight fusion algorithm. This achieves efficient integration and accurate analysis of multi-source data, outputting comprehensive monitoring results that include the spatial distribution of pollutants, spatiotemporal correlation characteristics, and traffic flow-pollution correlation. The dual-drive coupled prediction module innovatively integrates physical algorithms and AI residual correction, significantly improving the accuracy of pollutant propagation prediction. Simultaneously, the three-dimensional contribution tracing algorithm can accurately locate the source of pollution. This full-chain accurate sensing and intelligent prediction mechanism breaks through the limitations of traditional monitoring methods, enabling real-time and comprehensive understanding of the dynamics and causes of pollution within the tunnel. This provides solid data support for subsequent control decisions, effectively solving the problems of incomplete tunnel pollution monitoring and inaccurate prediction, and ensuring the forward-looking and targeted nature of air quality management.

[0027] Second, this invention constructs a three-dimensional evaluation index system covering air quality, energy consumption, and traffic efficiency through a multi-objective optimization decision-making module. Combined with a dynamic weight optimization algorithm to solve for the Pareto optimal control strategy set, it achieves a synergistic balance among multiple objectives. The closed-loop control mechanism formed by the equipment execution and feedback modules can dynamically adjust equipment operating parameters based on real-time control effects, ensuring the accurate implementation and continuous optimization of control commands. This multi-objective synergistic optimization and closed-loop control mode overcomes the drawbacks of traditional single-objective control. While ensuring that the air quality in the tunnel meets the standards, it effectively reduces equipment operating energy consumption, improves traffic efficiency and stability, and achieves an organic unity of environmental protection, energy conservation, and smooth traffic. It provides a new intelligent and efficient solution for highway tunnel operation and management, and promotes the transformation of tunnel ventilation and purification systems from passive response to proactive prediction and precise control.

[0028] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0030] Figure 1 A flowchart of an intelligent control system for ventilation and air purification in a highway tunnel;

[0031] Figure 2 This is a schematic diagram showing the connection relationships of various modules in an intelligent control system for ventilation and air purification in highway tunnels. Detailed Implementation

[0032] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0033] Example 1:

[0034] Operational scenario of a long urban highway tunnel during peak hours.

[0035] Multidimensional perception module:

[0036] An air parameter sensor group is deployed every 50 meters along the tunnel's longitudinal direction. In bidirectional tunnels, the groups are deployed separately for each direction of travel. Each group includes CO, NOx, and PM2.5 sensors, which can accurately capture changes in the concentration of key pollutants at different locations within the tunnel. Infrared thermal imagers are deployed at the tunnel entrance, exit, and every 1 km. Temperature and humidity sensors and visibility meters are deployed every 300 meters to monitor the physical state of the environment inside and outside the tunnel in real time, providing basic data for subsequent assessments of whether air quality meets standards. High-definition cameras and millimeter-wave radar combinations are deployed 500m before the tunnel entrance, 300m after the exit, and every 800m inside the tunnel to accurately collect traffic flow parameters, clarify vehicle traffic conditions, and aid in analyzing the correlation between traffic flow and pollution. A directional microphone is deployed every 200 meters along the tunnel sidewall as a sound wave sensor to promptly capture abnormal sounds, facilitating the rapid detection of potential safety hazards within the tunnel. Temperature sensors are installed at the motor windings of the ventilation equipment, vibration sensors at the bearings, and current sensors at the power supply of the purification equipment to monitor the equipment's operating status in real time and ensure stable operation. This module collects data on pollutant concentrations, traffic flow parameters, environmental physical parameters, abnormal noises, and equipment operating status within the tunnel. It also adds timestamps and spatial location information to all data to ensure traceability and provide accurate and comprehensive raw data support for subsequent module processing. Figure 1 As shown.

[0037] Multimodal fusion module:

[0038] After receiving various data output from the multi-dimensional sensing module, preprocessing is performed. Kalman filtering is used to denoise the air parameter sensor data, effectively reducing interference signals and improving the accuracy of pollutant concentration monitoring data. DB4 wavelet basis 3-layer decomposition denoising is used for the acoustic sensor data to reduce noise in abnormal sound data and ensure clear extraction of abnormal sound information. Gaussian filtering is used to denoise the image data, optimizing the quality of traffic flow image data and improving the accuracy of traffic flow parameter recognition. Based on timestamps, non-1Hz sampling data are unified to 1Hz through linear interpolation, ensuring synchronization of data from different sampling frequencies in the time dimension, facilitating subsequent data integration and analysis. Then, based on tunnel mileage marker coordinates, the data is mapped to a 5m×5m×3m spatial grid, achieving precise spatial correspondence and clearly presenting the monitoring status at different spatial locations within the tunnel. The 3σ criterion is used to remove outliers to avoid interference from abnormal data in subsequent analysis results. Missing data is filled using a distance-weighted average of adjacent time points at the same location, ensuring data integrity and laying a solid foundation for multi-source data fusion. After preprocessing, a dynamic weighted fusion algorithm is used to fuse the multi-source data. The expression for the dynamic weighted fusion algorithm is: ,in, It is a location ,time The concentration of pollutants after fusion It is spatial location. It was a moment. For the first Sensor-like sensors The monitoring value, No. Sensor-like position time Dynamic weights, It is the total number of sensor types. For multi-sensor data variance, As a consistency correction coefficient, it fully combines the advantages of various types of data and outputs comprehensive pollution monitoring results that include the spatial distribution of pollutant concentrations, spatiotemporal correlation characteristics, and traffic flow-pollution correlation, providing a comprehensive and reliable analytical basis for the dual-drive coupled prediction module.

[0039] Dual-drive coupling prediction module:

[0040] Based on the comprehensive pollution monitoring results output by the multimodal fusion module, a preliminary prediction is generated using the flow-diffusion coupling algorithm. The expression for the flow-diffusion coupling algorithm is as follows: ,in, For position ,time Preliminary predicted concentration, for time The initial concentration at that location, The segmented attenuation coefficient, The average velocity of the ventilation airflow. The pollutant diffusion coefficient, for time The intensity of pollution sources at the location The saturation time is affected by the pollution source. It is the pollution source impact coefficient. It is a natural constant. Based on spatial location, combined with the diffusion patterns of pollutants within the tunnel and the characteristics of airflow, a preliminary judgment is made regarding the future propagation trend of pollutants. Furthermore, by incorporating the AI ​​residual correction amount into a multi-dimensional dynamic equilibrium algorithm, the expression for the multi-dimensional dynamic equilibrium algorithm is: ,in, It is a moment and location The final predicted concentration of pollutants, It predicts the time step. The pollutant type sensitivity coefficient characterizes the degree to which different pollutants depend on the physical model. This serves as a spatial gradient adjustment factor, strengthening the constraints of the physical model in regions of abrupt concentration changes. It is a moment ,Location AI residual correction amount It is the confidence factor for AI correction. For position ,time The preliminary predicted concentration is then optimized and adjusted to reduce prediction errors and improve the accuracy of pollutant transmission prediction. Ultimately, a precise pollutant transmission prediction result is output, providing a scientific reference for the early formulation of pollution control strategies. Simultaneously, a three-dimensional contribution source tracing algorithm is used to locate the pollution source and clarify the contribution of various vehicle types to pollutant concentrations at different locations and times. The expression for the three-dimensional contribution source tracing algorithm is: ,in, It is a moment ,Location Place, No. The contribution of vehicle type to pollutant concentration is traced back to its source coefficient. It is the vehicle category number. For time window, It is the integral time variable. for time Place Traffic flow of similar vehicle types The time decay coefficient, It is a moment ,Location Place, No. Emission factors of this type of vehicle It is a moment ,Location The concentration of pollutants after fusion It is the first Traffic flow of similar vehicle types Is the predicted concentration of pollutants for the first Partial derivative of traffic flow for different vehicle types, It represents the pollutant concentration at x at time τ after multimodal fusion, helping staff to take targeted measures to reduce emissions from key pollution sources and control pollutant generation at the source.

[0041] Multi-objective optimization decision-making module:

[0042] Based on pollutant propagation predictions and real-time tunnel traffic conditions, a three-dimensional evaluation index system is constructed, encompassing air quality, energy consumption, and traffic efficiency. Regarding air quality, the system ensures that the average CO concentration every 5 minutes is ≤100ppm, NOx concentration is ≤5ppm, PM2.5 concentration is ≤0.05mg / m³, visibility is ≥100m, and the proportion of areas exceeding pollutant concentration thresholds within the total tunnel length is ≤5%, with the duration of exceeding thresholds being ≤3 minutes. This ensures that the air quality within the tunnel meets safe passage standards and protects the health of drivers and passengers. The energy consumption index covers equipment operating energy consumption and start-up / shutdown energy consumption. Operating energy consumption is calculated based on the actual operating parameters of axial flow fans, jet fans, electrostatic precipitators, and activated carbon adsorption devices. Start-up / shutdown energy consumption is calculated per fan start-up. The instantaneous energy consumption is calculated by adding 1.2 times the rated power and 0.8 times the rated power for each power switch of the purification equipment. The total energy consumption statistical period is 1 hour. Through precise energy consumption calculation, the energy consumption of the equipment is reduced as much as possible while meeting air quality requirements, thus reducing energy waste. The traffic efficiency indicators require that the average traffic flow in the tunnel per hour be ≥80% of the design flow, the average vehicle speed be ≥60km / h, the vehicle speed fluctuation coefficient be ≤0.2, the number of congestion events per hour be ≤1, and the average following distance of vehicles be ≥50m, to ensure smooth traffic in the tunnel, reduce traffic congestion, and improve traffic efficiency. A dynamic coupling optimization algorithm is used to solve for the Pareto optimal control strategy set. The expression of the dynamic coupling optimization algorithm is: ,in, It is a multi-objective comprehensive optimization index. As a normalized index for air quality, As a normalized energy consumption index, This is a normalized indicator for traffic efficiency. The energy consumption coupling coefficient is... As a traffic efficiency coupling coefficient, it fully balances the relationship between air quality, energy consumption and traffic efficiency, and selects a matching strategy from the optimal control strategy set based on the real-time status of the tunnel, which is then transformed into specific control commands for ventilation equipment and air purification equipment to ensure the optimal equipment operation scheme.

[0043] Device execution and feedback module:

[0044] The system receives coordinated control commands from the multi-objective optimization decision-making module, driving the axial flow fans, jet fans, electrostatic precipitators, and activated carbon adsorption devices. The axial flow fans activate a specified number and position according to the commands, adjusting to corresponding speed levels to enhance airflow within the tunnel and accelerate pollutant removal. A specified number of jet fans activate, adjusting their directional angle and speed to optimize airflow direction and improve ventilation. The electrostatic precipitators activate a specified number of stages and set operating power to effectively remove particulate pollutants from the air. The activated carbon adsorption devices activate a corresponding number of stages, setting activated carbon replacement warning thresholds and operating power to adsorb harmful gases and further purify the air. Simultaneously, the system collects equipment operating parameters and adjusted air quality data, updates system parameters, and feeds them back to the multi-dimensional sensing module and the dual-drive coupled prediction module. This allows the system to promptly grasp the equipment's operating performance and changes in air quality, adjusting subsequent control strategies based on feedback information to form a closed-loop control system. This ensures that air quality, equipment energy consumption, and traffic efficiency within the tunnel are always at their optimal levels.

[0045] In summary, during peak hours in long urban highway tunnels, this intelligent control system comprehensively collects accurate data through a multi-dimensional sensing module, providing support for subsequent processing; the multi-modal fusion module outputs reliable comprehensive monitoring results through preprocessing and dynamic weight fusion algorithms; the dual-drive coupling prediction module achieves accurate pollution prediction and source tracing through flow expansion coupling, multi-dimensional dynamic balance, and three-dimensional contribution source tracing algorithms; the multi-objective optimization decision-making module constructs a three-dimensional indicator system and determines the optimal strategy using dynamic coupling optimization algorithms; and the equipment execution and feedback module drives equipment operation and forms a closed-loop control, effectively balancing air quality, energy consumption, and traffic efficiency, ensuring stable tunnel operation during peak hours.

[0046] Example 2:

[0047] Low traffic volume and sudden pollution scenarios in mountainous highway tunnels.

[0048] Multidimensional perception module:

[0049] An air parameter sensor group is deployed every 80 meters along the tunnel's longitudinal direction. In the bidirectional tunnel, the groups are deployed separately for each direction of travel. Each group is equipped with CO, NOx, and PM2.5 sensors, enabling accurate monitoring of pollutant concentrations within the tunnel even under low traffic conditions, and timely detection of pollutant trends during sudden pollution events. Infrared thermal imagers are deployed at the tunnel entrance, exit, and every 1 km. Temperature and humidity sensors and visibility meters are deployed every 300 meters to monitor environmental physical parameters within the tunnel in real time, allowing for rapid assessment of environmental changes during sudden pollution events and providing data for determining the extent of pollution impact. Additional sensors are deployed 500m before the tunnel entrance and 300m after the exit. A combination of high-definition cameras and millimeter-wave radar is deployed every 800 meters inside the tunnel to accurately collect traffic flow parameters under low traffic volume, clarify vehicle traffic conditions, and facilitate analysis to determine whether sudden pollution is related to specific vehicles. A directional microphone is placed every 200 meters along the tunnel sidewall to promptly capture abnormal sounds that may accompany sudden pollution, such as vehicle malfunction noises, aiding in rapid investigation of the cause of pollution. Temperature sensors are installed on the motor windings of ventilation equipment, vibration sensors on bearings, and current sensors on the power supply of purification equipment to monitor equipment operating status in real time, ensuring normal startup and stable operation of equipment during sudden pollution events. This module collects data on pollutant concentration, traffic flow parameters, environmental physical parameters, abnormal sounds, and equipment operating status within the tunnel. All data is appended with timestamps and spatial location information, providing comprehensive and traceable raw data for subsequent modules to respond quickly and handle accurately in the event of sudden pollution. Figure 2 As shown.

[0050] Multimodal fusion module:

[0051] After receiving data from the multi-dimensional sensing module, preprocessing is performed. Kalman filtering is used to denoise the air parameter sensor data, effectively filtering out data interference and accurately capturing details of concentration changes when sudden pollution causes rapid changes in pollutant concentration. DB4 wavelet-based 3-layer decomposition denoising is used for the acoustic sensor data, clearly extracting abnormal sound information that may accompany sudden pollution, providing clear clues for tracing the pollution source. Gaussian filtering is used to denoise the image data, optimizing the quality of traffic flow images under low traffic volume, accurately identifying vehicle information, and assisting in determining whether a specific vehicle caused the sudden pollution. Based on timestamps, non-1Hz sampling data is unified to 1Hz through linear interpolation, ensuring that different types of data are synchronized in time, facilitating rapid integration and analysis of data changes before and after the sudden pollution event. Data is mapped to a 5m×5m×3m spatial grid based on tunnel mileage marker coordinates, accurately locating the spatial position of the sudden pollution within the tunnel and clarifying the pollution diffusion range. Outliers were removed using the 3σ criterion to avoid the impact of individual abnormal data caused by sudden contamination on the analysis results. Missing data were imputed using a distance-weighted average of data from adjacent time points at the same location, ensuring data integrity and providing a reliable foundation for multi-source data fusion. After preprocessing, a dynamic weighted fusion algorithm was used to fuse the multi-source data. The expression for the dynamic weighted fusion algorithm is as follows: It fully integrates information such as pollutant concentration, environmental parameters, traffic flow and abnormal sounds, and outputs comprehensive pollution monitoring results including the spatial distribution of pollutant concentration, spatiotemporal correlation characteristics and traffic flow-pollution correlation, providing key basis for the dual-drive coupled prediction module to quickly formulate response strategies.

[0052] Dual-drive coupling prediction module:

[0053] Based on comprehensive pollution monitoring results, a preliminary prediction is generated using a flow-diffusion coupling algorithm. Combined with the topographical characteristics of mountain tunnels and airflow patterns, the diffusion speed and direction of sudden pollution events are quickly determined, providing a preliminary understanding of the future development trend of the pollution. The expression for the flow-diffusion coupling algorithm is: By combining a multidimensional dynamic equilibrium algorithm with AI residual correction, and considering the unique characteristics of sudden pollution events, the preliminary prediction results are precisely adjusted to improve prediction accuracy and obtain pollutant propagation prediction results. This allows for timely implementation of measures to control pollution spread. The expression for the multidimensional dynamic equilibrium algorithm is as follows: Simultaneously, a three-dimensional contribution source tracing algorithm is used to locate the source of the sudden pollution. The expression for the three-dimensional contribution source tracing algorithm is: It can quickly determine the specific location of pollution and the main vehicle models contributing to it, helping staff to quickly pinpoint the source of pollution and take targeted measures to prevent the pollution from worsening.

[0054] Multi-objective optimization decision-making module:

[0055] Based on pollutant propagation predictions and real-time tunnel traffic conditions, a three-dimensional evaluation index system encompassing air quality, energy consumption, and traffic efficiency is constructed. Air quality must meet the following requirements: average CO concentration ≤100ppm, NOx concentration ≤5ppm, PM2.5 concentration ≤0.05mg / m³ every 5 minutes; visibility ≥100m; and the percentage of areas exceeding pollutant concentration thresholds ≤5% for a duration ≤3 minutes. In the event of sudden pollution, priority is given to restoring tunnel air quality to safe standards to ensure the safety of drivers and passengers. Energy consumption is calculated based on equipment operating energy consumption and start-up / shutdown energy consumption. Operating energy consumption is calculated according to the actual operating parameters of various equipment, while start-up / shutdown energy consumption is calculated according to relevant rules. Total energy consumption is statistically analyzed hourly. While rapidly controlling sudden pollution, equipment energy consumption is rationally controlled to avoid unnecessary energy consumption. Regarding traffic efficiency, the following requirements are ensured: average hourly traffic flow ≥80% of design flow, average vehicle speed ≥60km / h, speed fluctuation coefficient ≤0.2, ≤1 congestion event per hour, and average following distance ≥50m. In handling sudden pollution, the impact on tunnel traffic is minimized to maintain normal traffic order. The Pareto optimal control policy set is solved using a dynamic coupling optimization algorithm. The expression for the dynamic coupling optimization algorithm is as follows: By comprehensively balancing the handling of sudden pollution, energy consumption control, and traffic efficiency, and combining the real-time status of the tunnel to select matching strategies, these strategies are transformed into equipment control commands to ensure rapid and efficient handling of sudden pollution.

[0056] Device execution and feedback module

[0057] Upon receiving control commands, the system drives axial flow fans, jet fans, electrostatic precipitators, and activated carbon adsorption devices. It adjusts parameters such as the number, position, power, and angle of these devices according to the commands, for example, increasing the number of axial flow fans, increasing the speed of jet fans, and increasing the operating power of electrostatic precipitators and activated carbon adsorption devices. This rapidly enhances ventilation and purification capabilities, accelerates pollutant discharge and purification, and quickly restores air quality within the tunnel. The system collects equipment operating parameters and adjusted air quality data to monitor equipment operation and pollution control effectiveness in real time. It updates system parameters and feeds them back to the multi-dimensional sensing module and the dual-drive coupled prediction module, allowing the system to adjust control strategies promptly based on feedback. For example, as air quality gradually improves, it appropriately adjusts equipment operating parameters to reduce energy consumption, forming a closed-loop control system. This effectively handles sudden pollution events while balancing energy consumption and traffic efficiency, ensuring the overall stable operation of the tunnel.

[0058] In summary, for sudden pollution scenarios with low traffic volume in mountainous highway tunnels, the system's multi-dimensional sensing module promptly captures pollution and equipment data; the multi-modal fusion module preprocesses the data and outputs comprehensive monitoring results using a dynamic weighted fusion algorithm; the dual-drive coupled prediction module quickly predicts pollution propagation and locates the source; the multi-objective optimization decision-making module constructs a three-dimensional indicator system and formulates the optimal strategy using a dynamic coupled optimization algorithm; and the equipment execution and feedback module efficiently drives equipment to purify pollution, while simultaneously providing feedback data to form a closed loop. This approach rapidly handles sudden pollution while balancing energy consumption and traffic efficiency, ensuring the safe operation of the tunnel.

[0059] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An intelligent control system for ventilation and air purification in highway tunnels, characterized in that, The system includes: Multi-dimensional sensing module: Collects data on pollutant concentration, traffic flow parameters, environmental physical parameters, abnormal sounds, and equipment operating status in the tunnel through a distributed sensor network, and adds timestamps and spatial positioning information to all data; Multimodal fusion module: Receives data output from the multidimensional sensing module, and after preprocessing such as noise reduction, spatiotemporal alignment and outlier removal, it uses a dynamic weighted fusion algorithm to fuse multi-source data and outputs comprehensive pollution monitoring results including spatial distribution of pollutant concentration, spatiotemporal correlation characteristics and traffic flow-pollution correlation. Dual-drive coupled prediction module: Based on comprehensive pollution monitoring results, it generates preliminary predictions through flow-diffusion coupling algorithm, combines multi-dimensional dynamic balance algorithm with AI residual correction, and outputs pollutant propagation prediction results; at the same time, it uses three-dimensional contribution tracing algorithm to locate pollution sources; Multi-objective optimization decision module: Based on the pollutant propagation prediction results and the real-time traffic status of the tunnel, a three-dimensional evaluation index system including air quality, energy consumption and traffic efficiency is constructed. The dynamic weight optimization algorithm is used to solve the Pareto optimal control strategy set. The matching strategy is selected from the optimal control strategy set in combination with the real-time tunnel status and transformed into specific control instructions for ventilation equipment and air purification equipment. Equipment execution and feedback module: Receives coordinated control commands, drives the axial flow fan, jet fan, electrostatic dust removal, and activated carbon adsorption device to operate; collects equipment operating parameters and air quality data after adjustment, updates system parameters, and feeds them back to the multi-dimensional sensing module and the dual-drive coupled prediction module to form a closed-loop control.

2. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, The multi-dimensional sensing module includes a distributed sensor network comprising: an air parameter sensor group, with one group deployed every 50-80 meters along the longitudinal direction of the tunnel, and two groups deployed separately for each direction in a two-way tunnel; each group containing a CO sensor, a NOx sensor, and a PM2.5 sensor; an environmental status sensor group, with infrared thermal imagers deployed at the tunnel entrance, exit, and every 1 km, and temperature and humidity sensors and visibility meters deployed every 300 meters; a traffic sensing equipment group, with a combination of high-definition cameras and millimeter-wave radar deployed 500m before the tunnel entrance, 300m after the exit, and every 800m inside the tunnel; a sound wave sensor, with a directional microphone deployed every 200 meters along the tunnel sidewall; and an equipment status sensor group, with temperature sensors installed at the motor windings of the ventilation equipment, vibration sensors installed at the bearings, and current sensors installed at the power supply of the purification equipment.

3. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, The preprocessing steps in the multimodal fusion module include: using Kalman filtering to reduce noise in air parameter sensor data, using db4 wavelet basis 3-level decomposition to reduce noise in acoustic sensor data, and using Gaussian filtering to reduce noise in image data; unifying non-1Hz sampling data to 1Hz through linear interpolation based on timestamps; mapping data to a 5m×5m×3m spatial grid based on tunnel mileage marker coordinates; removing outliers using the 3σ criterion; and filling missing data with a distance-weighted average of adjacent time points at the same location.

4. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, In the multimodal fusion module, the expression for the dynamic weight fusion algorithm is: ,in, It is a location ,time The concentration of pollutants after fusion It is spatial location. It was a moment. For the first Sensor-like sensors The monitoring value, No. Sensor-like position time Dynamic weights, It is the total number of sensor types. For multi-sensor data variance, This is the consistency correction factor.

5. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, In the dual-drive coupling prediction module, the expression for the flow-expanding coupling algorithm is: ,in, For position ,time Preliminary predicted concentration, for time The initial concentration at that location, The segmented attenuation coefficient, The average velocity of the ventilation airflow. The pollutant diffusion coefficient, for time The intensity of pollution sources at the location The saturation time is affected by the pollution source. It is the pollution source impact coefficient. It is a natural constant. It refers to spatial location.

6. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, In the dual-drive coupling prediction module, the expression for the multidimensional dynamic equilibrium algorithm is: ,in, It is a moment and location The final predicted concentration of pollutants, It predicts the time step. The pollutant type sensitivity coefficient characterizes the degree to which different pollutants depend on the physical model. This serves as a spatial gradient adjustment factor, strengthening the constraints of the physical model in regions of abrupt concentration changes. It is a moment ,Location AI residual correction amount It is the confidence factor for AI correction. For position ,time Preliminary predicted concentration.

7. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, In the dual-drive coupling prediction module, the expression for the three-dimensional contribution tracing algorithm is: ,in, It is a moment ,Location Place, No. The contribution of vehicle type to pollutant concentration is traced back to its source coefficient. It is the vehicle category number. For time window, It is the integral time variable. for time Place Traffic flow of similar vehicle types The time decay coefficient, It is a moment ,Location Place, No. Emission factors of this type of vehicle It is a moment ,Location The concentration of pollutants after fusion It is the first Traffic flow of similar vehicle types Is the predicted concentration of pollutants for the first Partial derivative of traffic flow for different vehicle types, It is the pollutant concentration at x at time τ after multimodal fusion.

8. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, The multi-objective optimization decision-making module includes a three-dimensional evaluation index system that specifically comprises: Air quality indicators include hard constraints on pollutant concentration and spatial distribution constraints. The hard constraints are: average CO concentration ≤100ppm, NOx concentration ≤5ppm, PM2.5 concentration ≤0.05mg / m³ every 5 minutes, and visibility ≥100m. The spatial distribution constraints are: the proportion of areas with pollutant concentrations exceeding the threshold to the total tunnel length ≤5%, and the duration of exceeding the threshold ≤3 minutes. Energy consumption indicators include equipment operating energy consumption and start-up and shutdown energy consumption. Operating energy consumption is calculated based on the actual operating parameters of axial flow fans, jet fans, electrostatic dust removal devices, and activated carbon adsorption devices. Start-up and shutdown energy consumption is calculated based on the instantaneous energy consumption of 1.2 times the rated power added each time the fan is started and the instantaneous energy consumption of 0.8 times the rated power added each time the purification equipment switches power. The total energy consumption statistics period is 1 hour. Traffic efficiency indicators include traffic flow efficiency and driving stability. Traffic flow efficiency is defined as an average traffic volume in the tunnel of ≥80% of the design flow per hour and an average vehicle speed of ≥60km / h. Driving stability is defined as a speed fluctuation coefficient of ≤0.2, a number of congestion events of ≤1 per hour, and an average following distance of ≥50m.

9. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, In the multi-objective optimization decision module, the expression for the dynamically coupled optimization algorithm is: ,in, It is a multi-objective comprehensive optimization index. As a normalized index for air quality, As a normalized energy consumption index, This is a normalized indicator for traffic efficiency. The energy consumption coupling coefficient is... This is the traffic efficiency coupling coefficient.

10. The intelligent control system for ventilation and air purification in highway tunnels according to claim 1, characterized in that, The control parameters included in the Pareto optimal control strategy set in the device execution and feedback module are as follows: Ventilation equipment control parameters: number of axial flow fans, opening position and speed rating; number of jet fans, directional angle and speed rating; Air purification equipment control parameters: number of opening sections and operating power of electrostatic dust removal device; number of opening sections, activated carbon replacement warning threshold and operating power of activated carbon adsorption device.

Images