A method and system for online monitoring of environmental pollution

Abstract

The present invention relates to the field of environmental pollution monitoring, and specifically to an online environmental pollution monitoring method and system. The method collects pollutant concentration data of a parking lot and obtains pollutant concentration preprocessing data through data preprocessing; extracts spatial distribution characteristics and time characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to a multivariate spatiotemporal prediction model, and comprehensively analyzes the spatiotemporal trend of pollutant diffusion to obtain the diffusion trend of pollutant concentration and the predicted pollutant concentration; performs data analysis on the predicted pollutant concentration and a preset safety concentration threshold, and when the predicted pollutant concentration is greater than the preset safety concentration threshold, triggers an alarm and performs ventilation operation; solves the problem that pollutant monitoring in a parking lot is not comprehensive enough, pollutant concentration cannot be predicted in a timely and effective manner, and high-concentration areas of pollutants cannot be determined, resulting in the inability to timely optimize and manage air quality and protect personnel health.

Description

Technical Field

[0001] The invention relates to the field of environmental pollution monitoring, and in particular to an online environmental pollution monitoring method and system. Background Art

[0002] With the acceleration of urbanization and the rapid increase in the number of motor vehicles, the air pollution problem in parking lots, especially underground parking lots, is becoming increasingly serious. Due to its closed or semi-closed space structure, the ventilation conditions in underground parking lots are relatively poor, and pollutants (such as carbon monoxide, nitrogen oxides, particulate matter, etc.) are easily retained. Especially during peak parking hours and when the air flow is poor, exhaust pollution will accumulate significantly, posing a threat to the air quality in the parking lot and people's health.

[0003] Current parking lot air quality monitoring methods mostly rely on single-point monitoring data, which fails to fully reflect the distribution of pollutants in different areas. Especially in large underground parking lots, a single monitoring point is difficult to cover the pollution situation of the entire space. At the same time, the prediction ability of traditional monitoring systems is limited, and the real-time prediction of pollutant diffusion trends is insufficient, and it is impossible to effectively determine the high-concentration retention points that may appear in a short period of time. This deficiency will affect the timeliness and accuracy of the early warning system, thereby limiting the effectiveness of corresponding ventilation and control measures.

[0004] Therefore, there is an urgent need for an online environmental pollution monitoring method to achieve real-time monitoring of air pollution in parking lots, thereby optimizing air quality management and protecting personnel health. Summary of the invention

[0005] (1) Technical issues to be resolved

[0006] The purpose of the present invention is to provide an online monitoring method and system for environmental pollution, so as to solve the problem that the monitoring of pollutants in parking lots is not comprehensive enough, the concentration of pollutants cannot be predicted in a timely and effective manner, and the high concentration areas of pollutants cannot be determined, resulting in the system being unable to optimize the management of air quality in a timely manner and protect the health of personnel.

[0007] (2) Technical solution

[0008] To achieve the above object, on the one hand, the present invention provides an online monitoring method for environmental pollution, the method comprising:

[0009] Step 1: collect pollutant concentration data in the air inside the parking lot in real time through sensors deployed in the parking lot area, and obtain pollutant concentration preprocessed data through data preprocessing; the data preprocessing includes data filtering, outlier removal and time series synchronization processing.

[0010] Step 2: construct a multivariate spatiotemporal prediction model through convolutional neural networks and long short-term memory networks, extract the spatial distribution characteristics and time characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model, and conduct a comprehensive analysis of the spatiotemporal trend of pollutant diffusion to obtain the diffusion trend of pollutant concentration, and obtain the predicted pollutant concentration based on the diffusion trend.

[0011] Step 3: Perform data analysis on the predicted pollutant concentration and the preset safety concentration threshold. When the predicted pollutant concentration is greater than the preset safety concentration threshold, an alarm is triggered and ventilation operation is performed.

[0012] Furthermore, in step 2, the method of extracting spatial distribution characteristics and temporal characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model includes:

[0013] The parking lot area is divided into a preset number of grid units, and the grid units where sensors are deployed are recorded as monitoring grid units; the pollutant concentration data of the monitoring grid units are obtained, and the pollutant concentration data of all grid units are obtained by the Kriging interpolation method, and the pollutant concentration data of all grid units are input into the convolutional neural network, and a pollutant characteristic image is generated through the feature extraction layer of the convolutional neural network, and the spatial distribution data of the pollutant concentration is generated according to the pollutant characteristic image, and the spatial distribution data of the pollutant concentration is the spatial distribution characteristics of the pollutant diffusion.

[0014] The pollutant concentration data of the monitoring grid unit in multiple time periods are obtained to form a time series data set; the time series data set is used to generate a time series model of pollutant diffusion through a long short-term memory network; the pollutant concentration in future time periods is predicted through the time series model to obtain a change trend of the pollutant concentration, which is the time characteristic of pollutant diffusion.

[0015] Furthermore, in step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations includes:

[0016] The airflow parameters measured in the parking lot area are obtained, and the airflow parameters are simulated by simulation software to generate simulated airflow simulation data; the simulated airflow simulation data is standardized into a grid format to obtain standard airflow simulation data; the airflow parameters include wind speed, wind direction, and turbulence intensity.

[0017] The pollutant concentration data of all grid cells at the current moment are obtained and recorded as the first pollutant concentration data. The first pollutant concentration data and the standard airflow simulation data are calculated through the convection-diffusion equation to obtain the pollutant concentration prediction data of each grid cell at the next moment and recorded as the second pollutant concentration data.

[0018] The pollutant concentration data of all grid units at the next moment are obtained and recorded as the third pollutant concentration data, the third pollutant concentration data is matched with the second pollutant concentration data point by point, and the concentration difference of each grid unit is calculated, and the change trend of the pollutant concentration is analyzed according to the concentration difference; when the concentration difference is positive, the pollutant concentration in the grid unit is an increasing trend; when the concentration difference is negative, the pollutant concentration in the grid unit is a decreasing trend; when the concentration difference is positive and greater than a preset concentration difference threshold, the grid unit corresponding to the concentration difference is recorded as a pollutant significant change area; an initial pollutant diffusion path map is generated according to the pollutant significant change area, and the initial pollutant diffusion path map is subjected to spatial smoothing to obtain a final pollutant diffusion path map.

[0019] The pollutant final diffusion path map is used to identify the pollutant retention area in the parking lot area, and the retention area is marked as a retention point on the pollutant diffusion path map, and the retention point is used as the key monitoring object for ventilation operation; the retention area is a grid unit where the concentration difference is greater than a preset concentration difference threshold and the duration is greater than a preset safety time threshold.

[0020] Furthermore, in step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations further includes:

[0021] The historical pollutant concentration preprocessing data of the parking lot area is obtained, the historical pollutant concentration preprocessing data is divided according to a preset time window to obtain the first data, the first data is normalized to obtain the second data, the second data is analyzed by the Pearson correlation coefficient to obtain the correlation coefficient between the pollutants and generate a correlation coefficient matrix between the pollutants; the correlation coefficient is used to determine whether there is a synergistic diffusion effect between the pollutants.

[0022] The correlation coefficients in the correlation coefficient matrix are subjected to data analysis with the preset correlation threshold. When the correlation coefficient is greater than the preset correlation threshold, it is determined that there is a synergistic diffusion effect between the pollutants and they are recorded as high-correlation pollutants. When the concentrations of the high-correlation pollutants are greater than the preset safety concentration thresholds of the corresponding pollutants in the same time window, the alarm level is raised.

[0023] Furthermore, in step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations further includes:

[0024] The historical monitoring data of pollutant concentrations are obtained, and the historical correlation coefficient time series data of highly correlated pollutants are constructed in chronological order. The historical correlation coefficient time series data are trained through a long short-term memory network to obtain a correlation coefficient prediction model.

[0025] The correlation coefficient data of pollutant concentration in the most recent preset time window is obtained, and the correlation coefficient data is predicted by a correlation coefficient prediction model to obtain correlation coefficient prediction results for multiple preset prediction time steps in the future; the correlation coefficient prediction results are analyzed by data to obtain a correlation coefficient change trend curve. When the change trend curve shows an increasing trend and the correlation coefficient prediction results are all greater than the preset correlation threshold, it is judged that there is a significant synergistic diffusion effect between the pollutants and recorded as a synergistic diffusion risk; when the monitoring shows that there is a synergistic diffusion risk, the ventilation operation is pre-started.

[0026] Based on the same inventive concept, the present invention also provides an online environmental pollution monitoring system, the system comprising: a pollutant concentration data acquisition and preprocessing module, a pollutant predicted concentration prediction module, a pollutant predicted concentration determination and management module, and each module is sequentially connected in communication;

[0027] The pollutant concentration data collection and preprocessing module is used to collect pollutant concentration data in the air inside the parking lot in real time through sensors deployed in the parking lot area, and obtain pollutant concentration preprocessed data through data preprocessing; the data preprocessing includes data filtering, outlier removal and time series synchronization processing.

[0028] The pollutant concentration prediction module is used to construct a multivariate spatiotemporal prediction model through a convolutional neural network and a long short-term memory network, extract the spatial distribution characteristics and time characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model, and conduct a comprehensive analysis of the spatiotemporal trend of pollutant diffusion to obtain the diffusion trend of pollutant concentration, and obtain the predicted pollutant concentration based on the diffusion trend.

[0029] The pollutant predicted concentration determination and management module is used to perform data analysis on the pollutant predicted concentration and the preset safety concentration threshold. When the pollutant predicted concentration is greater than the preset safety concentration threshold, an alarm is triggered and ventilation operation is performed.

[0030] Furthermore, the pollutant concentration prediction module includes:

[0031] The spatial distribution characteristic analysis module is used to divide the parking lot area into a preset number of grid units, record the grid units where sensors are deployed as monitoring grid units; obtain the pollutant concentration data of the monitoring grid units and obtain the pollutant concentration data of all grid units through the Kriging interpolation method, input the pollutant concentration data of all grid units into the convolutional neural network, generate a pollutant characteristic image through the feature extraction layer of the convolutional neural network, and generate spatial distribution data of pollutant concentration based on the pollutant characteristic image, which is the spatial distribution characteristics of pollutant diffusion.

[0032] The time characteristic analysis module is used to obtain the pollutant concentration data of the monitoring grid unit in multiple time periods and form a time series data set; generate a time series model of pollutant diffusion through the long short-term memory network of the time series data set; predict the pollutant concentration in the future time period through the time series model to obtain the change trend of the pollutant concentration, and the change trend of the pollutant concentration is the time characteristic of the pollutant diffusion.

[0033] Furthermore, the pollutant prediction concentration determination and management module also includes:

[0034] The standard airflow simulation data calculation module is used to obtain the airflow parameters measured in the parking lot area, simulate the airflow parameters through simulation software to generate simulated airflow simulation data; standardize the simulated airflow simulation data into a grid format to obtain standard airflow simulation data; the airflow parameters include wind speed, wind direction, and turbulence intensity.

[0035] The second pollutant concentration data acquisition module is used to obtain the pollutant concentration data of all grid cells at the current moment and record it as the first pollutant concentration data, and calculate the first pollutant concentration data and the standard airflow simulation data through the convection-diffusion equation to obtain the pollutant concentration prediction data of each grid cell at the next moment and record it as the second pollutant concentration data.

[0036] The pollutant diffusion path map generation module is used to obtain the pollutant concentration data of all grid units at the next moment and record it as the third pollutant concentration data, match the third pollutant concentration data with the second pollutant concentration data point by point and calculate the concentration difference of each grid unit, and analyze the change trend of the pollutant concentration according to the concentration difference; when the concentration difference is a positive value, the pollutant concentration in the grid unit is an increasing trend; when the concentration difference is a negative value, the pollutant concentration in the grid unit is a decreasing trend; when the concentration difference is a positive value and greater than a preset concentration difference threshold, the grid unit corresponding to the concentration difference is recorded as a pollutant significant change area; generate an initial pollutant diffusion path map based on the pollutant significant change area, and obtain a final pollutant diffusion path map by spatially smoothing the initial pollutant diffusion path map.

[0037] The retention area management module is used to identify the retention area of ​​pollutants in the parking lot area through the pollutant final diffusion path map, mark the retention area as a retention point on the pollutant diffusion path map, and use the retention point as a key monitoring object for ventilation operation; the retention area is a grid unit whose concentration difference is greater than a preset concentration difference threshold and whose duration is greater than a preset safety time threshold.

[0038] Furthermore, the pollutant prediction concentration determination and management module also includes:

[0039] The correlation coefficient matrix calculation module is used to obtain the historical pollutant concentration preprocessing data of the parking lot area, divide the historical pollutant concentration preprocessing data according to a preset time window to obtain the first data, normalize the first data to obtain the second data, analyze the correlation coefficient between the pollutants through the Pearson correlation coefficient of the second data and generate a correlation coefficient matrix between the pollutants; the correlation coefficient is used to determine whether there is a synergistic diffusion effect between the pollutants.

[0040] The alarm level management module is used to perform data analysis on the correlation coefficients in the correlation coefficient matrix and the preset correlation threshold. When the correlation coefficient is greater than the preset correlation threshold, it is judged that there is a synergistic diffusion effect between the pollutants and they are recorded as high-correlation pollutants. When the concentrations of the high-correlation pollutants are greater than the preset safety concentration thresholds of the corresponding pollutants in the same time window, the alarm level is raised.

[0041] Furthermore, the pollutant prediction concentration determination and management module also includes:

[0042] The correlation coefficient prediction model acquisition module is used to obtain historical monitoring data of pollutant concentrations, and construct historical correlation coefficient time series data of highly correlated pollutants in chronological order, and obtain a correlation coefficient prediction model by training the historical correlation coefficient time series data through a long short-term memory network.

[0043] The synergistic diffusion risk determination module is used to obtain the correlation coefficient data of the pollutant concentration within the most recent preset time window, and predict the correlation coefficient data through the correlation coefficient prediction model to obtain the correlation coefficient prediction results for multiple preset prediction time steps in the future; the correlation coefficient prediction results are analyzed through data to obtain a correlation coefficient change trend curve. When the change trend curve shows an increasing trend and the correlation coefficient prediction results are all greater than the preset correlation threshold, it is judged that there is a significant synergistic diffusion effect between the pollutants and recorded as a synergistic diffusion risk; when the monitoring shows that there is a synergistic diffusion risk, the ventilation operation is pre-started.

[0044] (3) Beneficial effects

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. By combining the multivariate spatiotemporal prediction model of convolutional neural network and long short-term memory network, the diffusion trend of pollutant concentration in the parking lot area can be accurately predicted, the potential risk of exceeding the pollutant concentration can be discovered in advance, and the alarm can be automatically triggered when the pollutant concentration exceeds the preset safety threshold, so as to carry out ventilation operations in time, reduce the threat of air pollution to personnel health, and effectively optimize the environmental management of the parking lot area.

[0047] 2. Analyze the synergistic diffusion effect between pollutants through the Pearson correlation coefficient, and judge the relationship between pollutants based on the correlation coefficient matrix. When the concentration of highly correlated pollutants exceeds the safety threshold, the system will automatically increase the alarm level and adjust the ventilation operation.

[0048] 3. Combine historical monitoring data with the prediction model of long-short-term memory networks to identify the risk of coordinated diffusion in advance, and start ventilation operations in advance when the coordinated diffusion effect occurs, effectively improving the ability to control and prevent pollutant diffusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flowchart of an online environmental pollution monitoring method according to Example 1 of the present invention.

[0050] Figure 2 This is a schematic diagram of the module composition of an online environmental pollution monitoring system according to Example 2 of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Before giving examples, it is necessary to explain the application scenarios of the present invention. The present invention is an online monitoring method and system for environmental pollution. In the parking lot of a large underground shopping mall, the frequent entry and exit of vehicles causes a large amount of exhaust pollutants to be emitted. Especially during weekends and holidays, the number of vehicles surges, and the air quality inside the parking lot is greatly challenged. Due to the closed structure of the underground space and poor air circulation, pollutants (such as CO, NO2 and PM2.5) are easily retained in local areas, forming high pollution concentration areas, affecting the health of customers and staff.

[0053] Example 1: Figure 1As shown, this embodiment provides an online monitoring method for environmental pollution, the method comprising:

[0054] Step 1: The pollutant concentration data in the air inside the parking lot is collected in real time through the sensors arranged in the parking lot area, and the pollutant concentration data is obtained by data preprocessing to obtain pollutant concentration preprocessing data; the data preprocessing includes data filtering, outlier removal and time series synchronization processing; the sensors arranged in the parking lot area are used to monitor the concentration of pollutants (such as carbon monoxide, nitrogen dioxide, PM2.5, etc.) in the air in real time, and transmit these data to the parking lot air quality monitoring system. Since the data collected by the sensor in real time may contain noise, errors or outliers, data preprocessing is required to ensure the accuracy and consistency of the data. Data filtering removes noise by applying specific algorithms (such as low-pass filtering, mean filtering, etc.) to ensure the stability of the data, and is used to eliminate random fluctuations or instantaneous mutations, thereby improving the reliability of the data. Some abnormal data points may be generated during the sensor acquisition process (for example, extreme data caused by sensor failure or sudden changes in the environment). These abnormal values ​​can be identified and removed through outlier detection (such as Z-score, IQR method, etc.) to avoid affecting subsequent analysis. Each sensor may have different sampling times or frequencies. Through time series synchronization processing, the data is aligned to a unified time axis to facilitate comprehensive analysis of the sensor data and ensure spatiotemporal consistency. This ensures that monitoring data at different locations and times can be calculated collaboratively to improve the accuracy of the analysis.

[0055] Step 2, constructing a multivariate spatiotemporal prediction model through convolutional neural networks and long short-term memory networks, extracting the spatial distribution characteristics and temporal characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model, and comprehensively analyzing the spatiotemporal trend of pollutant diffusion to obtain the diffusion trend of pollutant concentration, and obtaining the predicted pollutant concentration based on the diffusion trend; the multivariate spatiotemporal prediction model is used to process the spatial distribution and temporal evolution of pollutant concentration. Through the joint analysis of the spatial and temporal characteristics of pollutant concentration, the multivariate spatiotemporal prediction model can obtain the comprehensive diffusion trend of pollutants, thereby reflecting the changes in pollutant concentration with location and time, and predicting the pollutant concentration in future time periods and locations based on the comprehensive spatiotemporal trend of the spatiotemporal prediction model.

[0056] Step 3, the predicted concentration of the pollutant is analyzed with the preset safety concentration threshold. When the predicted concentration of the pollutant is greater than the preset safety concentration threshold, an alarm is triggered and ventilation operation is performed. In order to ensure that the air quality in the parking lot area meets the health standards, the system will set a safety concentration threshold. The safety concentration threshold is based on environmental regulations, health standards or industry regulations. Exceeding the maximum allowable concentration value may endanger human health. Data analysis compares the predicted concentration of pollutants with the preset safety concentration threshold. When the predicted concentration is greater than the preset safety concentration threshold, it means that the concentration of pollutants in the parking lot has exceeded the safety limit, which may cause the air quality to deteriorate and endanger human health. The system will automatically trigger an alarm. The alarm may notify the management personnel in the form of sound and light alarms, or send an alarm message through the network so that timely measures can be taken. In addition to the alarm, the system will also start the ventilation equipment for air circulation according to the preset control strategy to reduce the concentration of pollutants and keep the air quality within a safe range. The ventilation operation is to introduce fresh air from the outside or accelerate the flow of internal air to help dilute and discharge harmful pollutants, so as to ensure that the air quality in the parking lot is always at a healthy and safe level.

[0057] Furthermore, in step 2, the method of extracting spatial distribution characteristics and temporal characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model includes:

[0058] The parking lot area is divided into a preset number of grid units, and the grid units where sensors are deployed are recorded as monitoring grid units; the pollutant concentration data of the monitoring grid units are obtained and the pollutant concentration data of all grid units are obtained by the Kriging interpolation method, and the pollutant concentration data of all grid units are input into the convolutional neural network, and the pollutant characteristic image is generated by the feature extraction layer of the convolutional neural network. The spatial distribution data of the pollutant concentration is generated according to the pollutant characteristic image, and the spatial distribution data of the pollutant concentration is the spatial distribution characteristics of the pollutant diffusion; the parking lot area is divided into multiple grid units, and each grid unit represents a spatial area in the parking lot. The purpose of dividing the grid units is to accurately monitor the concentration changes of pollutants and ensure that the spatial distribution of air quality can be captured at the micro level. For example, the parking lot can be divided into several small rectangular or square areas (grid units), and sensors can be deployed in some key grid units to collect pollutant concentration data in the air in real time. For grid units without sensors (unmonitored grid units), the Kriging interpolation method is used to predict their pollutant concentrations. The Kriging interpolation method is a spatial interpolation method that uses the pollutant concentration data and geographic spatial relationship of the monitoring grid units to interpolate and predict the unmonitored grid units to obtain the pollutant concentration data of all grid units. Convolutional neural network (CNN) is a deep learning algorithm used for image processing and feature extraction. Pollutant concentration data is input into the convolutional neural network (CNN) model. The pollutant concentration data is usually represented in the form of a matrix, where each matrix unit corresponds to the pollutant concentration value of a grid unit. CNN has multiple layers, among which the "convolution layer" can be used to extract features from the input data, automatically learn and extract features in the spatial distribution (for example, the law of concentration distribution, hot spots, etc.) based on the pollutant concentration data. CNN converts the spatial information of pollutant concentration into a pollutant characteristic image, that is, an image that can reflect the spatial distribution pattern of pollutant concentration. The pollutant characteristic image can represent the change of pollutant concentration in the grid unit, reflecting the distribution characteristics of pollutants in space. Based on the output pollutant characteristic image of the convolutional neural network, the system further generates spatial distribution data of pollutant concentration, describing the distribution of pollutant concentration in different grid units. The final generated spatial distribution data of pollutant concentration reflects the diffusion pattern and concentration distribution of pollutants in the parking lot area. For example, some areas may have high pollutant concentrations (such as near the lanes), while other areas may have low pollutant concentrations (such as areas far away from emission sources).

[0059] The pollutant concentration data of the monitoring grid unit in multiple time periods are obtained and a time series data set is formed; the time series data set is used to generate a time series model of pollutant diffusion through a long short-term memory network; the pollutant concentration in the future time period is predicted by the time series model to obtain the change trend of the pollutant concentration, and the change trend of the pollutant concentration is the time characteristic of pollutant diffusion. In the monitoring grid unit of the parking lot area, the sensor will record the pollutant concentration data in multiple time periods. For example, the pollutant concentration is recorded every few minutes or hours to form a time series. These data constitute the change of pollutant concentration in the grid unit over a period of time. The pollutant concentration data recorded over time are arranged in chronological order to form a time series data set. The data set records the complete trajectory of how the pollutant concentration changes over time. The formation of the time series data set is the basis for predicting the future trend of pollutants because it contains the historical change information of the pollutant concentration over time. The long short-term memory network is a deep learning model suitable for processing time series data, which can capture the time dependency of data during the processing process. LSTM is good at remembering long-term related information in the time series, such as the periodicity or sudden increase of pollutant concentration changes. The formed time series data set is input into the LSTM model. LSTM generates a time series model that describes the change of pollutant concentration by learning the time dependency in the data. It is used to capture the temporal dynamics of pollutant diffusion, such as the concentration change in the morning and evening. The generated time series model can be used to predict the change of pollutant concentration in the future. The trend of pollutant concentration change in the future can be obtained through LSTM prediction, indicating whether the pollutants are rising, stable or falling. For example, a period of dense vehicle traffic may lead to an increase in pollutant concentration, while a period of ventilation or a decrease in vehicle traffic may lead to a decrease in concentration.

[0060] Furthermore, in step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations includes:

[0061] The airflow parameters measured in the parking lot area are obtained, and the airflow parameters are simulated by simulation software to generate simulated airflow simulation data; the simulated airflow simulation data is standardized into a grid format to obtain standard airflow simulation data; the airflow parameters include wind speed, wind direction, and turbulence intensity; the simulation software will simulate the flow of airflow in the parking lot in a virtual environment according to parameters such as wind speed, wind direction, and turbulence intensity. This embodiment uses computational fluid dynamics (CFD) simulation software. CFD simulation software is the most commonly used tool for airflow simulation. It simulates the movement of fluids, including air, gas, etc., through numerical methods. For problems such as pollutant diffusion and air circulation, CFD software can accurately simulate and predict the behavior of airflow. The airflow simulation data obtained by simulation can describe the flow pattern of air in the parking lot in detail, such as the difference in flow velocity of airflow in different areas, the bending and turning of airflow in space, etc. The simulated airflow data is standardized so that the data has a consistent scale and unit. The airflow data is formatted according to the grid unit, and each grid unit has a set of corresponding airflow parameter data, so as to ensure that the spatial position and data format of the airflow parameters are consistent, and the obtained standard airflow simulation data can clearly represent the state of airflow in different grid units, thereby supporting subsequent pollutant diffusion analysis.

[0062] Obtain pollutant concentration data of all grid cells at the current moment and record it as first pollutant concentration data, calculate the first pollutant concentration data and standard airflow simulation data through the convection-diffusion equation to obtain pollutant concentration prediction data of each grid cell at the next moment and record it as second pollutant concentration data;

[0063] The pollutant concentration data of all grid units at the next moment are obtained and recorded as the third pollutant concentration data, the third pollutant concentration data is matched point by point with the second pollutant concentration data, and the concentration difference of each grid unit is calculated, and the change trend of the pollutant concentration is analyzed according to the concentration difference; when the concentration difference is positive, the pollutant concentration in the grid unit is an increasing trend; when the concentration difference is negative, the pollutant concentration in the grid unit is a decreasing trend; when the concentration difference is positive and greater than the preset concentration difference threshold, the grid unit corresponding to the concentration difference is recorded as a pollutant significant change area; according to the pollutant significant change area, the pollutant initial diffusion path map is generated, and the pollutant initial diffusion path map is obtained by spatial smoothing to obtain the pollutant final diffusion path map; a positive concentration difference indicates an increase in pollutant concentration, which may be that the pollutant is diffusing to the area or the pollutant suddenly increases in emission, and the pollutant concentration is on an upward trend. A negative concentration difference indicates a decrease in pollutant concentration, which may be that the pollutant has been diluted or the pollutant emission has decreased, and the pollutant concentration is on a downward trend. The preset concentration difference threshold is set to ensure that only areas with large changes are identified without being disturbed by small fluctuations. Areas with significant changes in pollutants may be hot spots for pollutant diffusion or sources of concentration growth, and deserve special attention and monitoring. By identifying areas with significant changes in pollutants, an initial pollutant diffusion path map can be generated, that is, the path of pollutants diffusing from the significant change area, which is used to present the direction and path of increasing pollutant concentrations and help determine the trend and direction of pollutant diffusion. Spatial smoothing can generate a smoother and more accurate final pollutant diffusion path map, which can more realistically reflect the diffusion process of pollutants in the parking lot. Spatial smoothing is used to eliminate some local noise data, making the path map smoother, continuous, and more practical. Smoothing is usually done using mathematical methods (such as Gaussian filtering, interpolation, etc.).

[0064] The pollutant retention area in the parking lot area is identified by the pollutant final diffusion path map, and the retention area is marked as a retention point on the pollutant diffusion path map, and the retention point is used as the key monitoring object of the ventilation operation; the retention area is a grid unit whose concentration difference is greater than the preset concentration difference threshold and the duration is greater than the preset safety time threshold. In the process of pollutant diffusion, some areas may cause pollutants to gather in specific areas due to local changes in airflow (such as low wind speed, no obvious wind direction change, etc.), which is difficult to be effectively discharged or diffused, that is, the retention area. The preset concentration difference threshold is preset according to the actual situation. For example, if the pollutant concentration in some areas is higher than that in other areas by a certain degree (for example, more than 10% or 20%), it can be regarded as an indication of the retention area. The abnormal pollutant concentration in the retention area must not only reach the preset concentration difference threshold at a certain moment, but also maintain this abnormal state for a period of time. This is to avoid occasional small fluctuations or short-term airflow changes from being misjudged as retention phenomena. The preset safety time threshold is used to determine whether the retention phenomenon lasts long enough, indicating that the pollutants in the area are really retained and cannot be effectively diffused. For example, if the pollutant concentration of a grid unit is higher than the preset concentration difference threshold and lasts for more than 10 minutes, the area can be identified as a retention area. On the pollutant diffusion path map, based on the judgment results of the preset concentration difference threshold and the preset safety time threshold, the system will mark the location of these retention areas on the diffusion path map as "retention points". These retention points are areas where pollutants are concentrated or retained in the parking lot area. These areas may become potential dangerous points with excessively high pollutant concentrations. Retention points may have a significant impact on air quality due to abnormally high pollutant concentrations. In order to avoid further accumulation of pollutants in the retention areas, the system will mark these retention points as key monitoring objects. The marking and monitoring of retention points allow the ventilation system to be adjusted according to actual conditions, such as increasing air circulation in areas near retention points, or concentrating the discharge of pollutants to ensure that pollutants in these areas can be discharged and treated in a timely manner.

[0065] Furthermore, in step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations further includes:

[0066] The historical pollutant concentration preprocessing data of the parking lot area is obtained, the historical pollutant concentration preprocessing data is divided according to the preset time window to obtain the first data, the first data is normalized to obtain the second data, the second data is analyzed by the Pearson correlation coefficient to obtain the correlation coefficient between the pollutants and generate the correlation coefficient matrix between the pollutants; the correlation coefficient is used to determine whether there is a synergistic diffusion effect between the pollutants; the time window refers to the division of historical data according to a certain time period. For example, by dividing the data according to different time units such as minutes and hours, the system can analyze the changes in pollutant concentrations in each time period. The first data is the original data set obtained after the division, which usually contains pollutant concentration information in different time windows. Normalization is to unify the dimensions of the data, and usually converts the numerical range of different data to the same standard range (such as 0 to 1) so that the concentration data of different pollutants are comparable in subsequent analysis. The purpose of normalization is to eliminate the impact of dimensional differences so that different dimensions of the data have the same scale, which is convenient for subsequent analysis and modeling. The Pearson correlation coefficient is a statistical indicator that measures the strength of the linear relationship between two variables. Its value ranges from -1 to 1: 1 indicates a perfect positive correlation, that is, an increase in the concentration of one pollutant is accompanied by an increase in the concentration of another pollutant. -1 indicates a perfect negative correlation, that is, an increase in the concentration of one pollutant is accompanied by a decrease in the concentration of another pollutant. 0 indicates no linear relationship. The correlation coefficient matrix is ​​a square matrix. Each element in the matrix represents the Pearson correlation coefficient between two pollutants, which can fully reflect the relationship between each pollutant. By calculating the Pearson correlation coefficient between pollutant concentrations, the correlation of concentration changes of different pollutants in the same time period can be analyzed. If the correlation coefficient of two pollutants is very high, it means that there is a certain degree of synergistic diffusion effect between them. The synergistic diffusion effect refers to the fact that the concentration changes of two or more pollutants in space or time are synchronized. They may be affected by the same factors together, or there is a certain physical or chemical relationship between them. The existence of the synergistic diffusion effect means that the changes in pollutant concentrations do not occur independently, but may be driven by a common source, diffusion mechanism or other external factors.

[0067] The correlation coefficient in the correlation coefficient matrix is ​​analyzed with the preset correlation threshold. When the correlation coefficient is greater than the preset correlation threshold, it is judged that there is a synergistic diffusion effect between the pollutants and they are recorded as highly correlated pollutants. When the concentration of the highly correlated pollutants is greater than the preset safety concentration threshold of the corresponding pollutants in the same time window, the alarm level is increased. The preset correlation threshold is a pre-set standard used to judge whether the correlation between pollutants is strong enough to be considered to have a synergistic diffusion effect. For example, if the correlation coefficient of two pollutants is greater than the threshold (such as 0.8), it is considered that the concentration changes of the two pollutants are highly synchronous and there may be a synergistic diffusion relationship. The synergistic diffusion effect means that the concentration changes of multiple pollutants in the same time period show similar trends, which may be caused by common factors (such as airflow, climatic conditions, emission sources, etc.). If the correlation coefficient between two pollutants is greater than the preset correlation threshold (such as 0.8), then the two pollutants are considered to be "highly correlated pollutants". The synergistic diffusion effect between these pollutants may indicate that they are affected by the same environmental factors at the same time, or that there is a chemical reaction or physical transmission relationship between the pollutants. For example, the safety concentration of a pollutant may be set to 50 units. If its concentration exceeds 50 units, an early warning will be triggered. The concentration of highly correlated pollutants refers to whether the concentrations of pollutants that are considered to have a synergistic diffusion effect exceed their respective safety concentration thresholds within the same time window. If their concentrations exceed the safety concentration thresholds, it means that there is a more serious pollution risk. When the concentrations of highly correlated pollutants exceed their respective safety concentration thresholds at the same time, it means that the pollutant concentration may reach a dangerous or abnormal level, and the alarm level needs to be raised. Raising the alarm level means initiating more intensive emergency response measures, which means stricter monitoring, faster response or warnings to personnel.

[0068] Furthermore, in step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations further includes:

[0069] Obtain historical monitoring data on pollutant concentrations, and construct historical correlation coefficient time series data for highly correlated pollutants in chronological order. Train the historical correlation coefficient time series data through a long short-term memory network to obtain a correlation coefficient prediction model; arrange the correlation coefficients of each highly correlated pollutant in chronological order to form a time series data set, which can show how the correlation between pollutants changes over time. Through these historical data, it is possible to capture the changing trend of the relationship between pollutants over time. The model obtained through LSTM training can predict the changes in the correlation coefficients between pollutants at a certain moment or over a period of time in the future, which can help identify whether a synergistic diffusion effect will occur between pollutants, and can also predict whether the strength of the association between pollutants has changed.

[0070] The correlation coefficient data of pollutant concentration in the most recent preset time window is obtained, and the correlation coefficient data is predicted by the correlation coefficient prediction model to obtain the correlation coefficient prediction results of multiple preset prediction time steps in the future; the correlation coefficient prediction results are analyzed to obtain the correlation coefficient change trend curve, when the change trend curve shows an increasing trend and the correlation coefficient prediction results are all greater than the preset correlation threshold, it is judged that there is a significant synergistic diffusion effect between pollutants and recorded as synergistic diffusion risk; when the monitoring shows that there is a synergistic diffusion risk, the ventilation operation is pre-started. "Time window" refers to the pollutant concentration data in the most recent period of time. For example, the data of the past minutes and hours can be selected as the time window. The prediction time step refers to the specific step length of the model to predict the correlation coefficient in the future period of time. For example, the correlation coefficient trend of pollutants in the next 10 minutes, 30 minutes or one hour is predicted. The model predicts future changes based on historical data and the existing correlation of pollutant concentrations. The predicted correlation coefficient results are matched with time to construct a change trend curve to reflect the change pattern of the correlation between pollutants. The increase and decrease trend of the correlation of pollutants can be intuitively displayed, and the slope and shape of the curve can be analyzed to determine whether the correlation between pollutants is enhanced or weakened. The trend curve can help determine whether pollutants are beginning to show strong synergistic effects. If the trend curve shows that the correlation between pollutants is continuing to grow, it means that the concentration changes of pollutants are becoming more and more synchronized, which may indicate that the synergistic diffusion effect is strengthening. If the predicted correlation coefficient value continues to be higher than a preset threshold (for example, 0.7), it indicates that there is a strong synergistic effect between pollutants. The risk of synergistic diffusion means that pollutants will further aggregate in space, causing a sharp increase in local pollution concentrations. When the system detects the risk of synergistic diffusion, that is, the correlation between pollutants increases significantly and the risk of concentration diffusion is high, in order to avoid further increases in pollution concentrations and impacts on the environment, the system will start pre-start ventilation operations. Pre-start means ventilation control in advance to mitigate the possible impact of pollutant concentrations. For example, the pollutant concentration can be reduced by adjusting the exhaust airflow of the ventilation system or introducing fresh air from outside before a sharp increase in pollution concentration is detected.

[0071] Embodiment 2: Based on the same inventive concept, this embodiment also provides an online environmental pollution monitoring system, the system comprising: a pollutant concentration data acquisition and preprocessing module, a pollutant predicted concentration prediction module, a pollutant predicted concentration determination and management module, and each module is sequentially connected in communication;

[0072] The pollutant concentration data collection and preprocessing module is used to collect pollutant concentration data in the air inside the parking lot in real time through sensors deployed in the parking lot area, and obtain pollutant concentration preprocessed data through data preprocessing; the data preprocessing includes data filtering, outlier removal and time series synchronization processing.

[0073] The pollutant concentration prediction module is used to construct a multivariate spatiotemporal prediction model through a convolutional neural network and a long short-term memory network, extract the spatial distribution characteristics and time characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model, and conduct a comprehensive analysis of the spatiotemporal trend of pollutant diffusion to obtain the diffusion trend of pollutant concentration, and obtain the predicted pollutant concentration based on the diffusion trend.

[0074] The pollutant predicted concentration determination and management module is used to perform data analysis on the pollutant predicted concentration and the preset safety concentration threshold. When the pollutant predicted concentration is greater than the preset safety concentration threshold, an alarm is triggered and ventilation operation is performed.

[0075] The pollutant concentration prediction module includes:

[0076] The spatial distribution characteristic analysis module is used to divide the parking lot area into a preset number of grid units, record the grid units where sensors are deployed as monitoring grid units; obtain the pollutant concentration data of the monitoring grid units and obtain the pollutant concentration data of all grid units through the Kriging interpolation method, input the pollutant concentration data of all grid units into the convolutional neural network, generate a pollutant characteristic image through the feature extraction layer of the convolutional neural network, and generate spatial distribution data of pollutant concentration based on the pollutant characteristic image, which is the spatial distribution characteristics of pollutant diffusion.

[0077] The time characteristic analysis module is used to obtain the pollutant concentration data of the monitoring grid unit in multiple time periods and form a time series data set; generate a time series model of pollutant diffusion through the long short-term memory network of the time series data set; predict the pollutant concentration in the future time period through the time series model to obtain the change trend of the pollutant concentration, and the change trend of the pollutant concentration is the time characteristic of the pollutant diffusion.

[0078] The pollutant prediction concentration determination and management module also includes:

[0079] The standard airflow simulation data calculation module is used to obtain the airflow parameters measured in the parking lot area, simulate the airflow parameters through simulation software to generate simulated airflow simulation data; standardize the simulated airflow simulation data into a grid format to obtain standard airflow simulation data; the airflow parameters include wind speed, wind direction, and turbulence intensity.

[0080] The second pollutant concentration data acquisition module is used to obtain the pollutant concentration data of all grid cells at the current moment and record it as the first pollutant concentration data, and calculate the first pollutant concentration data and the standard airflow simulation data through the convection-diffusion equation to obtain the pollutant concentration prediction data of each grid cell at the next moment and record it as the second pollutant concentration data.

[0081] The pollutant diffusion path map generation module is used to obtain the pollutant concentration data of all grid units at the next moment and record it as the third pollutant concentration data, match the third pollutant concentration data with the second pollutant concentration data point by point and calculate the concentration difference of each grid unit, and analyze the change trend of the pollutant concentration according to the concentration difference; when the concentration difference is a positive value, the pollutant concentration in the grid unit is an increasing trend; when the concentration difference is a negative value, the pollutant concentration in the grid unit is a decreasing trend; when the concentration difference is a positive value and greater than a preset concentration difference threshold, the grid unit corresponding to the concentration difference is recorded as a pollutant significant change area; generate an initial pollutant diffusion path map based on the pollutant significant change area, and obtain a final pollutant diffusion path map by spatially smoothing the initial pollutant diffusion path map.

[0082] The retention area management module is used to identify the retention area of ​​pollutants in the parking lot area through the pollutant final diffusion path map, mark the retention area as a retention point on the pollutant diffusion path map, and use the retention point as a key monitoring object for ventilation operation; the retention area is a grid unit whose concentration difference is greater than a preset concentration difference threshold and whose duration is greater than a preset safety time threshold.

[0083] The pollutant prediction concentration determination and management module also includes:

[0084] A correlation coefficient matrix calculation module is used to obtain historical pollutant concentration preprocessing data of the parking lot area, divide the historical pollutant concentration preprocessing data according to a preset time window to obtain first data, normalize the first data to obtain second data, and analyze the correlation coefficients between pollutants using the Pearson correlation coefficient of the second data to generate a correlation coefficient matrix between pollutants; the correlation coefficient is used to determine whether there is a synergistic diffusion effect between pollutants;

[0085] The alarm level management module is used to perform data analysis on the correlation coefficients in the correlation coefficient matrix and the preset correlation threshold. When the correlation coefficient is greater than the preset correlation threshold, it is judged that there is a synergistic diffusion effect between the pollutants and they are recorded as high-correlation pollutants. When the concentrations of the high-correlation pollutants are greater than the preset safety concentration thresholds of the corresponding pollutants in the same time window, the alarm level is raised.

[0086] The pollutant prediction concentration determination and management module also includes:

[0087] A correlation coefficient prediction model acquisition module is used to obtain historical monitoring data of pollutant concentrations, and construct historical correlation coefficient time series data of highly correlated pollutants in chronological order, and obtain a correlation coefficient prediction model by training the historical correlation coefficient time series data through a long short-term memory network;

[0088] The synergistic diffusion risk determination module is used to obtain the correlation coefficient data of the pollutant concentration within the most recent preset time window, and predict the correlation coefficient data through the correlation coefficient prediction model to obtain the correlation coefficient prediction results for multiple preset prediction time steps in the future; the correlation coefficient prediction results are analyzed through data to obtain a correlation coefficient change trend curve. When the change trend curve shows an increasing trend and the correlation coefficient prediction results are all greater than the preset correlation threshold, it is judged that there is a significant synergistic diffusion effect between the pollutants and recorded as a synergistic diffusion risk; when the monitoring shows that there is a synergistic diffusion risk, the ventilation operation is pre-started.

[0089] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0090] Finally, it should be noted that: Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for online monitoring of environmental pollution, characterized in that: The method comprises: Step 1: collect pollutant concentration data in the air inside the parking lot in real time through sensors deployed in the parking lot area, and obtain pollutant concentration preprocessing data through data preprocessing; the data preprocessing includes data filtering, outlier removal and time series synchronization processing; the pollutant concentration data is uniformly aligned on the time axis through time series synchronization processing, so as to facilitate the comprehensive analysis of the data from each sensor and ensure the consistency of time and space; Step 2, constructing a multivariate spatiotemporal prediction model through a convolutional neural network and a long short-term memory network, extracting the spatial distribution characteristics and time characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model, and performing a comprehensive analysis on the spatiotemporal trend of pollutant diffusion to obtain the diffusion trend of pollutant concentration, and obtaining the predicted pollutant concentration according to the diffusion trend; Step 3, performing data analysis on the predicted pollutant concentration and the preset safety concentration threshold. When the predicted pollutant concentration is greater than the preset safety concentration threshold, an alarm is triggered and ventilation operation is performed; the ventilation operation is performed by introducing fresh air from the outside or accelerating the flow of internal air; In step 2, the method of extracting the spatial distribution characteristics and time characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model includes: The parking lot area is divided into a preset number of grid units, and the grid units where sensors are deployed are recorded as monitoring grid units; the pollutant concentration data of the monitoring grid units are obtained and the pollutant concentration data of all grid units are obtained by the Kriging interpolation method, and the pollutant concentration data of all grid units are input into the convolutional neural network, and a pollutant characteristic image is generated through the feature extraction layer of the convolutional neural network, and the spatial distribution data of the pollutant concentration is generated according to the pollutant characteristic image, and the spatial distribution data of the pollutant concentration is the spatial distribution characteristics of the pollutant diffusion; Acquire pollutant concentration data of the monitoring grid unit in multiple time periods and form a time series data set; generate a time series model of pollutant diffusion through the long short-term memory network; predict the pollutant concentration in the future time period through the time series model to obtain the change trend of the pollutant concentration, which is the time characteristic of pollutant diffusion; In step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations include: Obtain airflow parameters measured in the parking lot area, and simulate the airflow parameters through simulation software to generate simulated airflow simulation data; standardize the simulated airflow simulation data into a grid format to obtain standard airflow simulation data; the airflow parameters include wind speed, wind direction, and turbulence intensity; Obtain pollutant concentration data of all grid cells at the current moment and record it as first pollutant concentration data, calculate the first pollutant concentration data and standard airflow simulation data through the convection-diffusion equation to obtain pollutant concentration prediction data of each grid cell at the next moment and record it as second pollutant concentration data; Obtain pollutant concentration data of all grid cells at the next moment and record them as third pollutant concentration data, match the third pollutant concentration data with the second pollutant concentration data point by point and calculate the concentration difference of each grid cell, and analyze the change trend of pollutant concentration according to the concentration difference; when the concentration difference is a positive value, the pollutant concentration in the grid cell is on an increasing trend; when the concentration difference is a negative value, the pollutant concentration in the grid cell is on a decreasing trend; when the concentration difference is a positive value and is greater than a preset concentration difference threshold, the grid cell corresponding to the concentration difference is recorded as a pollutant significant change area; generate an initial pollutant diffusion path map according to the pollutant significant change area, and obtain a final pollutant diffusion path map by spatially smoothing the initial pollutant diffusion path map; The pollutant retention area in the parking lot area is identified by the pollutant final diffusion path map, and the retention area is marked as a retention point on the pollutant diffusion path map, and the retention point is used as a key monitoring object for ventilation operation; the retention area is a grid unit whose concentration difference is greater than a preset concentration difference threshold and whose duration is greater than a preset safety time threshold; In step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations further includes: Acquire historical pollutant concentration preprocessing data of the parking lot area, divide the historical pollutant concentration preprocessing data according to a preset time window to obtain first data, normalize the first data to obtain second data, analyze the correlation coefficient between pollutants using the Pearson correlation coefficient of the second data and generate a correlation coefficient matrix between pollutants; the correlation coefficient is used to determine whether there is a synergistic diffusion effect between pollutants; The correlation coefficients in the correlation coefficient matrix are analyzed with the preset correlation threshold. When the correlation coefficient is greater than the preset correlation threshold, it is judged that there is a synergistic diffusion effect between the pollutants and they are recorded as highly correlated pollutants. When the concentrations of the highly correlated pollutants are greater than the preset safety concentration thresholds of the corresponding pollutants in the same time window, the alarm level is increased. The correlation coefficient matrix is ​​a square matrix, and each element in the correlation coefficient matrix represents the Pearson correlation coefficient between two pollutants. In step 3, when the predicted pollutant concentration is greater than a preset safety concentration threshold, triggering an alarm and performing ventilation operations further includes: Obtain historical monitoring data of pollutant concentrations, and construct historical correlation coefficient time series data of highly correlated pollutants in chronological order, and obtain a correlation coefficient prediction model by training the historical correlation coefficient time series data through a long short-term memory network; The correlation coefficient data of pollutant concentration in the most recent preset time window is obtained, and the correlation coefficient data is predicted by a correlation coefficient prediction model to obtain correlation coefficient prediction results for multiple preset prediction time steps in the future; the correlation coefficient prediction results are analyzed by data to obtain a correlation coefficient change trend curve, and when the change trend curve shows an increasing trend and the correlation coefficient prediction results are all greater than the preset correlation threshold, it is judged that there is a significant synergistic diffusion effect between the pollutants and recorded as a synergistic diffusion risk; when the monitoring shows that there is a synergistic diffusion risk, the ventilation operation is pre-started; pre-starting means performing ventilation control in advance.

2. An online environmental pollution monitoring system, characterized in that: The system includes: a pollutant concentration data acquisition and preprocessing module, a pollutant predicted concentration prediction module, and a pollutant predicted concentration determination and management module, and each module is sequentially connected in communication; The pollutant concentration data collection and preprocessing module is used to collect pollutant concentration data in the air inside the parking lot in real time through sensors arranged in the parking lot area, and obtain pollutant concentration preprocessing data by data preprocessing; the data preprocessing includes data filtering, outlier removal and time series synchronization processing; the pollutant concentration data is uniformly aligned on the time axis through time series synchronization processing, so as to facilitate the comprehensive analysis of the data of each sensor and ensure the consistency of time and space; A pollutant concentration prediction module is used to construct a multivariate spatiotemporal prediction model through a convolutional neural network and a long short-term memory network, extract the spatial distribution characteristics and time characteristics of pollutant diffusion from the pollutant concentration preprocessing data according to the multivariate spatiotemporal prediction model, and perform a comprehensive analysis on the spatiotemporal trend of pollutant diffusion to obtain the diffusion trend of pollutant concentration, and obtain the predicted pollutant concentration according to the diffusion trend; The pollutant predicted concentration determination and management module is used to perform data analysis on the predicted pollutant concentration and the preset safety concentration threshold. When the predicted pollutant concentration is greater than the preset safety concentration threshold, an alarm is triggered and ventilation operation is performed; the ventilation operation is performed by introducing fresh air from the outside or accelerating the flow of internal air; The pollutant concentration prediction module includes: A spatial distribution characteristic analysis module is used to divide the parking lot area into a preset number of grid units, record the grid units where sensors are deployed as monitoring grid units; obtain pollutant concentration data of the monitoring grid units and obtain pollutant concentration data of all grid units through Kriging interpolation method, input the pollutant concentration data of all grid units into a convolutional neural network, generate a pollutant characteristic image through a feature extraction layer of the convolutional neural network, and generate spatial distribution data of pollutant concentration based on the pollutant characteristic image, wherein the spatial distribution data of pollutant concentration is the spatial distribution characteristics of pollutant diffusion; A time characteristic analysis module is used to obtain the pollutant concentration data of the monitoring grid unit in multiple time periods and form a time series data set; generate a time series model of pollutant diffusion through the long short-term memory network; predict the pollutant concentration in the future time period through the time series model to obtain the change trend of the pollutant concentration, and the change trend of the pollutant concentration is the time characteristic of the pollutant diffusion; The pollutant prediction concentration determination and management module includes: The standard airflow simulation data calculation module is used to obtain the airflow parameters measured in the parking lot area, simulate the airflow parameters through simulation software to generate simulated airflow simulation data; standardize the simulated airflow simulation data into a grid format to obtain standard airflow simulation data; the airflow parameters include wind speed, wind direction, and turbulence intensity; A second pollutant concentration data acquisition module is used to acquire the pollutant concentration data of all grid cells at the current moment and record it as the first pollutant concentration data, calculate the first pollutant concentration data and the standard airflow simulation data through the convection-diffusion equation to obtain the pollutant concentration prediction data of each grid cell at the next moment and record it as the second pollutant concentration data; A pollutant diffusion path map generation module is used to obtain the pollutant concentration data of all grid cells at the next moment and record it as the third pollutant concentration data, match the third pollutant concentration data with the second pollutant concentration data point by point and calculate the concentration difference of each grid cell, and analyze the change trend of the pollutant concentration according to the concentration difference; when the concentration difference is a positive value, the pollutant concentration in the grid cell is an increasing trend; when the concentration difference is a negative value, the pollutant concentration in the grid cell is a decreasing trend; when the concentration difference is a positive value and greater than a preset concentration difference threshold, the grid cell corresponding to the concentration difference is recorded as a pollutant significant change area; generate an initial pollutant diffusion path map according to the pollutant significant change area, and obtain a final pollutant diffusion path map by spatially smoothing the initial pollutant diffusion path map; A retention area management module is used to identify the retention area of ​​pollutants in the parking lot area through the pollutant final diffusion path map, mark the retention area as a retention point on the pollutant diffusion path map, and use the retention point as a key monitoring object for ventilation operation; the retention area is a grid unit where the concentration difference is greater than a preset concentration difference threshold and the duration is greater than a preset safety time threshold; The pollutant prediction concentration determination and management module also includes: A correlation coefficient matrix calculation module is used to obtain historical pollutant concentration preprocessing data of the parking lot area, divide the historical pollutant concentration preprocessing data according to a preset time window to obtain first data, normalize the first data to obtain second data, and analyze the correlation coefficients between pollutants using the Pearson correlation coefficient of the second data to generate a correlation coefficient matrix between pollutants; the correlation coefficient is used to determine whether there is a synergistic diffusion effect between pollutants; The alarm level management module is used to perform data analysis on the correlation coefficients in the correlation coefficient matrix and the preset correlation threshold. When the correlation coefficient is greater than the preset correlation threshold, it is judged that there is a synergistic diffusion effect between the pollutants and they are recorded as highly correlated pollutants. When the concentrations of the highly correlated pollutants are greater than the preset safety concentration thresholds of the corresponding pollutants in the same time window, the alarm level is increased. The correlation coefficient matrix is ​​a square matrix, and each element in the correlation coefficient matrix represents the Pearson correlation coefficient between two pollutants. The pollutant prediction concentration determination and management module also includes: A correlation coefficient prediction model acquisition module is used to obtain historical monitoring data of pollutant concentrations, and construct historical correlation coefficient time series data of highly correlated pollutants in chronological order, and obtain a correlation coefficient prediction model by training the historical correlation coefficient time series data through a long short-term memory network; The synergistic diffusion risk determination module is used to obtain the correlation coefficient data of the pollutant concentration within the most recent preset time window, and predict the correlation coefficient data through the correlation coefficient prediction model to obtain the correlation coefficient prediction results for multiple preset prediction time steps in the future; the correlation coefficient prediction results are analyzed through data to obtain a correlation coefficient change trend curve. When the change trend curve shows an increasing trend and the correlation coefficient prediction results are all greater than the preset correlation threshold, it is judged that there is a significant synergistic diffusion effect between the pollutants and recorded as a synergistic diffusion risk; when the monitoring shows that there is a synergistic diffusion risk, the ventilation operation is pre-started; pre-starting means performing ventilation control in advance.

Images