Industrial VOCS emission quality evaluation method and system based on big data

Abstract

The invention discloses an industrial VOCS emission quality evaluation method and system based on big data, and the method comprises the steps: calculating the emission amount in unit time based on the VOCS concentration and flow, obtaining the emission intensity characteristics, calculating the treatment efficiency characteristics through comparing the VOCS concentration before and after treatment, and simulating the concentration distribution under a complex terrain through an ADMS, and obtaining the environmental influence characteristics; fusing the emission intensity feature, the treatment efficiency feature and the environmental influence feature through a cross-modal attention mechanism; an LSTM neural network is utilized to predict the VOCS emission trend in a future time period, and the operation efficiency of treatment equipment is quantified through a random forest algorithm; and inputting the fusion feature vector into the dynamic evaluation model, and outputting an industrial VOCS emission quality evaluation report. The limitation of traditional single index evaluation is avoided, and the evaluation result is more comprehensive.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of environmental monitoring and pollution control technology, in particular to an industrial VOCs emission quality evaluation method and system based on big data. BACKGROUND

[0002] Traditional VOCs emission evaluation methods mainly rely on fixed-point monitoring or periodic sampling analysis, which has problems such as high data lag, limited coverage, and difficulty in reflecting real-time emission dynamics. In addition, existing methods are based on a single data source and lack the ability to analyze multi-dimensional data, and lack real-time monitoring of VOCs treatment facility efficiency, resulting in low accuracy of evaluation results. SUMMARY

[0003] The present application is designed to solve the above problems, and provides an industrial VOCs emission quality evaluation method and system based on big data.

[0004] To achieve the above-mentioned purpose, the technical scheme of the present application is as follows: further, in the above-mentioned industrial VOCs emission quality evaluation method based on big data, the industrial VOCs emission quality evaluation method comprises the following steps: Obtain multi-source industrial emission data, preprocess the multi-source industrial emission data to obtain standard industrial emission data; Calculate the emission intensity feature based on the VOCs concentration and flow rate, calculate the treatment efficiency feature by comparing the VOCs concentration before and after treatment, use AERMOD to quickly predict the diffusion range and use ADMS to simulate the concentration distribution under complex terrain to obtain the environmental impact feature; Fuse the emission intensity feature, treatment efficiency feature and environmental impact feature through a cross-modal attention mechanism to obtain a fusion feature vector; Use an LSTM neural network to predict the VOCs emission trend in the future period, use a random forest algorithm to quantify the operation efficiency of the treatment equipment, and use a Transformer model to evaluate the health risk index of VOCs emission to establish a dynamic evaluation model; Input the fusion feature vector into the dynamic evaluation model to output an industrial VOCs emission quality evaluation report.

[0005] Further, in the above-mentioned industrial VOCs emission quality evaluation method based on big data, the step of obtaining multi-source industrial emission data and preprocessing the multi-source industrial emission data to obtain standard industrial emission data comprises: Obtain emission source data, equipment operating condition data, environmental meteorological data and production process data to obtain multi-source industrial emission data; The missing values ​​in the multi-source industrial emission data were processed using the mean imputation method to obtain the cleaning industry emission data; The multi-source data of different magnitudes in the cleaning industry emission data are converted into the same magnitude using the Z-score standardization method and then integrated to obtain standard industrial emission data.

[0006] Furthermore, in the aforementioned big data-based industrial VOCs emission quality assessment method, the emission intensity characteristics are obtained by calculating the emission per unit time based on VOCs concentration and flow rate. The treatment efficiency characteristics are calculated by comparing VOCs concentrations before and after treatment. The diffusion range is quickly predicted using AERMOD, and the concentration distribution under complex terrain is simulated using ADMS to obtain environmental impact characteristics, including: The calculated emissions per unit time are organized and analyzed. By comparing the emissions under different time periods and different production conditions, the emission intensity characteristics are obtained. Subtracting the concentration after treatment from the concentration before treatment and comparing it with the concentration before treatment yields the characteristics of treatment efficiency. Obtain detailed terrain data, including the degree of terrain undulation, mountain range orientation, and river distribution, and input the terrain data, emission source data, and meteorological data into the ADMS model; Based on the influence of complex terrain on airflow, the concentration distribution of VOCs at different locations was simulated, and the environmental impact characteristics were obtained.

[0007] Furthermore, in the aforementioned big data-based industrial VOCs emission quality assessment method, the fusion of emission intensity characteristics, governance efficiency characteristics, and environmental impact characteristics through a cross-modal attention mechanism to obtain a fused feature vector includes: The query vector of emission intensity features is compared with the key vector of governance efficiency features, and the similarity between the two is calculated to obtain the attention weight of emission intensity features on governance efficiency features. Calculate the attention weight between emission intensity characteristics and environmental impact characteristics, and the attention weight between governance efficiency characteristics and environmental impact characteristics; Based on the calculated attention weights, the value vectors of each feature are summed using weighted averages. The value vector of the emission intensity feature is adjusted according to the corresponding attention weights, and the value vectors of the governance efficiency feature and the environmental impact feature are also adjusted to obtain the fused feature vector.

[0008] Furthermore, in the aforementioned big data-based industrial VOCs emission quality assessment method, the step of using an LSTM neural network to predict future VOCs emission trends, quantifying the operational efficiency of treatment equipment using a random forest algorithm, assessing the health risk index of VOCs emissions based on a Transformer model, and establishing a dynamic assessment model includes: The data is divided into training samples and test samples using a random forest algorithm, the operating parameters in the training samples are used as the input of the model, and the corresponding treatment effect is used as the target of model learning; The processed data is divided into a training part and a validation part, and the training part is used to learn the relationship between VOCs concentration, population exposure, VOCs toxicity factors and health risks.

[0009] Further, in the above-mentioned industrial VOCs emission quality evaluation method based on big data, the LSTM neural network is used to predict the VOCs emission trend in the future period, the random forest algorithm is used to quantify the operation efficiency of the treatment equipment, and the health risk index of VOCs emission is evaluated according to the Transformer model, and a dynamic evaluation model is established, and the method further comprises: The future VOCs emission trend predicted by the LSTM neural network, the operation efficiency of the treatment equipment quantified by the random forest algorithm, and the health risk index evaluated by the Transformer model are integrated to establish a dynamic evaluation model.

[0010] Further, in the above-mentioned industrial VOCs emission quality evaluation method based on big data, the fusion feature vector is input into the dynamic evaluation model, and an industrial VOCs emission quality evaluation report is output, comprising: The fusion feature vector is input into the dynamic evaluation model, and an evaluation result is output, the evaluation result is divided into different grades, at least including excellent, good, qualified and unqualified.

[0011] Further, in an industrial VOCs emission quality evaluation system based on big data, the industrial VOCs emission quality evaluation system comprises the following modules: A multi-source data acquisition module is used to acquire multi-source industrial emission data, and the multi-source industrial emission data is preprocessed to obtain standard industrial emission data; A multi-source feature extraction module is used to calculate the emission intensity feature based on VOCs concentration and flow rate, calculate the treatment efficiency feature by comparing VOCs concentration before and after treatment, use AERMOD to quickly predict the diffusion range and use ADMS to simulate the concentration distribution under complex terrain to obtain the environmental impact feature; A feature vector fusion module is used to fuse the emission intensity feature, the treatment efficiency feature and the environmental impact feature through a cross-modal attention mechanism to obtain a fusion feature vector; An evaluation model establishment module is used to predict the VOCs emission trend in the future period using an LSTM neural network, quantify the operation efficiency of the treatment equipment using a random forest algorithm, and evaluate the health risk index of VOCs emission according to a Transformer model to establish a dynamic evaluation model. The evaluation report generation module is configured to input the fusion feature vector into the dynamic evaluation model to output an industrial VOCS emission quality evaluation report.

[0012] Further, in the industrial VOCS emission quality evaluation system based on big data, the evaluation model establishment module comprises the following units: The input unit is configured to divide the data into training samples and test samples by using a random forest algorithm, use the operation parameters in the training samples as the input of the model, and use the corresponding governance effect as the target of model learning. The division unit is configured to divide the processed data into a training part and a verification part, and use the training part to learn the relationship between VOCS concentration, population exposure, VOCS toxicity factors and health risks.

[0013] Further, in the industrial VOCS emission quality evaluation system based on big data, the evaluation report generation module comprises the following units: The establishment unit is configured to integrate the future VOCS emission trend predicted by the LSTM neural network, the governance device operation efficiency quantified by the random forest algorithm, and the health risk index evaluated by the Transformer model to establish a dynamic evaluation model.

[0014] The beneficial effects are that 1. the industrial VOCS emission is comprehensively characterized from multiple dimensions. The emission intensity feature reflects the scale and intensity of the emission, the governance efficiency feature reflects the effectiveness of the governance measures, and the environmental impact feature shows the effect of the emission on the surrounding environment. This multi-dimensional evaluation method avoids the limitations of traditional single-index evaluation, and makes the evaluation results more comprehensively reflect the actual quality of industrial VOCS emission. 2. Potential risks are found in time, which provides a scientific basis for formulating prevention and response measures, and helps to control the emission risk at a low level. BRIEF DESCRIPTION OF DRAWINGS

[0015] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are for purposes of illustration only and are not considered a limitation of the present application.

[0016] Figure 1 The first embodiment of the industrial VOCS emission quality evaluation method based on big data in the embodiments of the present application is shown in the first embodiment schematic diagram. Figure 2 The second embodiment of the industrial VOCS emission quality evaluation method based on big data in the embodiments of the present application is shown in the second embodiment schematic diagram. Figure 3A first embodiment of a big data-based industrial VOCs emission quality evaluation system is shown in the embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the objects, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0018] Those skilled in the art can understand that, unless specifically stated, the singular forms "a", "an" and "the" used herein also include the plural forms. It should be further understood that the use of the word "comprise" in the specification of the present application means that the features, integers, steps, operations, elements and / or components exist, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0019] The present application is specifically described below with reference to the accompanying drawings, Figure 1 The industrial VOCs emission quality evaluation method based on big data comprises the following steps: Step 101, obtaining multi-source industrial emission data, pre-processing the multi-source industrial emission data to obtain standard industrial emission data; Specifically, in the present embodiment, the emission source data, equipment working condition data, environmental meteorological data and production process data are obtained to obtain the multi-source industrial emission data; The missing values in the multi-source industrial emission data are processed by using the mean filling method to obtain the cleaned industrial emission data; The multi-source data of different orders of magnitude in the cleaned industrial emission data are converted to the same order of magnitude by using the Z-score standardization method, and then integrated to obtain the standard industrial emission data.

[0020] Specifically; 1.1 Data acquisition; Emission source data: Obtain basic emission information of enterprises through enterprise pollution reporting system, and at the same time, real-time collect real-time data such as VOCs concentration and emission amount by means of online monitoring equipment installed at each emission port.

[0021] Equipment working condition data: Real-time monitor the running parameters of the equipment such as temperature, pressure, speed, etc. by using various sensors installed on the production equipment, and transmit the data to the data acquisition terminal.

[0022] Environmental meteorological data: Obtain conventional meteorological data from local meteorological stations, including air temperature, humidity, wind speed, wind direction, etc.; at the same time, obtain more detailed meteorological data such as atmospheric stability through professional meteorological data platform.

[0023] Production process data: export production process related data from enterprise production management system, such as raw material type and dosage, production cycle, production load, etc.

[0024] 1.2 Data preprocessing; Data cleaning: for missing values, if the missing proportion is small (less than 5%), the mean filling method is used for processing; if the missing proportion is large, interpolation filling is carried out combined with the change trend before and after the data. The 3σ principle is used, that is, if the data value deviates from the mean value more than 3 times the standard deviation, it is determined as an abnormal value and is removed.

[0025] Data standardization: convert multi-source data of different magnitudes to the same magnitude, use Z-score standardization method, formula: \(Z=(X-\mu) / \sigma\), where \(X\) is the original data, \(\mu\) is the mean, \(\sigma\) is the standard deviation, so that the data is converted to normal distribution data with mean value 0 and standard deviation 1.

[0026] Data integration: through data correlation technology, according to enterprise code, time stamp and other key information, integrate emission source data, equipment working condition data, environmental meteorological data and production process data into a unified data set to obtain standard industrial emission data.

[0027] Step 102, based on the concentration of VOCs and the flow rate, the emission intensity characteristics are obtained, the treatment efficiency characteristics are calculated by comparing the VOCs concentration before and after treatment, the diffusion range is quickly predicted by AERMOD, and the concentration distribution under complex terrain is simulated by ADMS to obtain environmental impact characteristics; Specifically, the unit time emission amount calculated in the embodiment is sorted and analyzed, and the emission intensity characteristics are obtained by comparing the emission amount under different time periods and different production conditions; The concentration before treatment is subtracted from the concentration after treatment, and compared with the concentration before treatment to obtain the treatment efficiency characteristics; Obtain detailed terrain data, including the degree of terrain undulation, mountain trend and river distribution, input the terrain data and emission source data, meteorological data into the ADMS model; According to the influence of complex terrain on air movement, the concentration distribution of VOCs at different positions is simulated to obtain the environmental impact characteristics.

[0028] Specifically; 2.1 Emission intensity characteristics; The emission amount per unit time is calculated based on the VOCs concentration and the flow, and the calculation formula is: emission amount per unit time = VOCs concentration x flow. The emission amount per unit time calculated is taken as the emission intensity feature, which can reflect the emission scale of VOCs per unit time.

[0029] 2.2 governance efficiency feature; The VOCs concentration data before and after treatment are collected respectively, and the governance efficiency is calculated according to the formula: governance efficiency = (pre-treatment concentration - post-treatment concentration) / pre-treatment concentration x 100%, which is taken as the governance efficiency feature. This feature reflects the reduction effect of the treatment measures on VOCs emission.

[0030] 2.3 environmental impact feature; The AERMOD model is used to input the emission source parameters (emission height, outlet diameter, emission temperature, etc.), meteorological parameters (wind speed, wind direction, atmospheric stability, etc.), and quickly predict the diffusion range of VOCs at different distances and regions.

[0031] For complex terrain areas, the ADMS model is used to import detailed terrain data (elevation, terrain slope, etc.), simulate the concentration distribution of VOCs under complex terrain, and obtain the environmental impact feature by combining the diffusion range and concentration distribution.

[0032] Step 103, fuse the emission intensity feature, the governance efficiency feature and the environmental impact feature through the cross-modal attention mechanism to obtain a fusion feature vector; Specifically, in this embodiment, the query vector of the emission intensity feature is compared with the key vector of the governance efficiency feature, and the similarity of the two is calculated to obtain the attention weight of the emission intensity feature on the governance efficiency feature. The attention weight between the emission intensity feature and the environmental impact feature, and the attention weight between the governance efficiency feature and the environmental impact feature are calculated. According to the calculated attention weights, the value vectors of each feature are weighted and summed, the value vector of the emission intensity feature is adjusted according to the corresponding attention weight, and the value vectors of the governance efficiency feature and the environmental impact feature are adjusted to obtain the fusion feature vector.

[0033] Specifically; 3.1 feature preprocessing; Before feature fusion, the emission intensity feature, the governance efficiency feature and the environmental impact feature need to be preprocessed to ensure consistency in data form and magnitude, laying a foundation for subsequent cross-modal attention mechanism processing. For the emission intensity feature, the data may include the emission amount per unit time in different time periods and the trend description of the emission intensity change. It is necessary to arrange these data into a unified structured form, such as arranging them in sequence according to time sequence, and at the same time, standardizing the emission amount values to eliminate the influence of different magnitudes; The governance efficiency feature usually involves specific numerical values of governance efficiency and changes under different working conditions. Similarly, these data need to be converted into standardized structured data, for example, classifying and arranging the governance efficiency values under different working conditions according to specific categories, and standardizing them so that the values are within the same interval range; The environmental impact feature is more complex, including geographical information description of the diffusion range, such as boundary coordinates of the affected area, sensitive areas involved, and concentration distribution data such as concentration values at different locations. The geographical information description needs to be converted into quantifiable structured data, such as representing different sensitive areas through coding; the concentration distribution data needs to be arranged and standardized to ensure that they are matched in magnitude with other features.

[0034] 3.2 Construction of cross-modal attention mechanism; The core of the cross-modal attention mechanism is to let different types of features pay attention to each other and identify the degree of association between them, so as to determine their importance in the fusion process; First, construct the corresponding query vector, key vector, and value vector for each feature. The query vector is used to find relevant information in other features; the key vector is used to match the query vector of other features to calculate the degree of association; the value vector is the core information in the feature for fusion; For the emission intensity feature, construct the corresponding query, key, and value vectors according to the emission scale and trend it reflects. The governance efficiency feature generates corresponding vectors based on the effectiveness and changes of governance. The environmental impact feature constructs vectors combining diffusion range and concentration distribution information to ensure that the vectors accurately represent the core content of the feature.

[0035] 3.3 Attention weight calculation; By calculating the similarity between the query vectors and key vectors of different features, the attention weight is determined, and the weight size reflects the importance of the feature in fusion; Compare the query vector of the emission intensity feature with the key vector of the governance efficiency feature to calculate their similarity and obtain the attention weight of the emission intensity feature on the governance efficiency feature. This weight reflects the degree of emphasis of the emission intensity feature on the governance efficiency feature; Similarly, the attention weight between the emission intensity feature and the environmental impact feature, and the attention weight between the governance efficiency feature and the environmental impact feature are calculated. In the calculation process, the physical meaning and actual correlation of various features are comprehensively considered, for example, when the emission intensity is large, the attention weight of the environmental impact feature may be relatively high.

[0036] 3.4 Feature fusion implementation According to the calculated attention weight, the value vectors of various features are weighted and summed to realize the fusion of features. The value vector of the emission intensity feature is adjusted according to its corresponding attention weight, and the value vectors of the governance efficiency feature and the environmental impact feature are also adjusted. Then the value vectors of the three adjusted features are added to obtain a preliminary fusion feature vector. In the fusion process, the important information of each feature is fully considered, and for those features with high correlation with other features and great influence on the overall evaluation, a higher weight is used to make them occupy a more important position in the fusion vector.

[0037] 3.5 Fusion feature vector optimization The preliminary fusion feature vector may have some redundant information or noise, which needs to be optimized to improve its accuracy and effectiveness. By analyzing the contribution degree of each element in the fusion feature vector, redundant elements with little influence on the evaluation result are removed, the dimension of the vector is reduced, and the key information is retained.

[0038] In addition, the fusion feature vector is subjected to consistency test to ensure that it can accurately reflect the characteristics of emission intensity, governance efficiency and environmental impact. If abnormal information inconsistent with the actual situation is found in the vector, the previous steps need to be traced back to check whether there is a problem in the feature preprocessing or attention weight calculation, and to be corrected until the ideal fusion feature vector is obtained.

[0039] Step 104, using LSTM neural network to predict the VOCS emission trend in future period, using random forest algorithm to quantify the running efficiency of the treatment equipment, using Transformer model to evaluate the health risk index of VOCS emission, and establishing a dynamic evaluation model; Specifically, in the present embodiment, the random forest algorithm is used to divide the data into training samples and test samples, and the running parameters in the training samples are used as the input of the model, and the corresponding treatment effect is used as the learning target of the model. The processed data is divided into training part and verification part, the training part is used to learn the relationship between VOCS concentration, population exposure, VOCS toxicity factor and health risk.

[0040] Integrate the future VOCs emission trend predicted by the LSTM neural network, the governance equipment operation efficiency quantified by the random forest algorithm, and the health risk index evaluated by the Transformer model to establish a dynamic evaluation model.

[0041] Specifically; 4.1 LSTM neural network predicts VOCs emission trend in future period; 4.1.1 Data preparation; Collect historical VOCs emission data, covering different production cycles, different seasons, and different production conditions. These data should include specific emission time and corresponding VOCs emission amount to ensure data integrity and continuity. At the same time, the data should be screened to remove abnormal emission data caused by equipment failure, sudden accidents, etc. to ensure that the data used for model training can truly reflect the emission law under normal production conditions.

[0042] 4.1.2 Model training and optimization; Divide the sorted historical data into training set and validation set according to time sequence. The training set is used to let the LSTM neural network learn the pattern of VOCs emission change over time, such as whether the emission increases when the production load increases, whether there is an emission peak in a certain season, etc. During the training process, the model parameters such as network layers and hidden unit number are adjusted to improve the fitting degree of the model to the historical data. The validation set is used to test the prediction effect of the model, and the model parameters are further optimized according to the validation results until the model can accurately reproduce the historical emission trend.

[0043] 4.1.3 Emission trend prediction; When the model is trained, input the recent VOCs emission data, which will be used as the starting point for model prediction. The LSTM neural network will learn the historical emission pattern and combine the recent data trend to predict the VOCs emission in different future periods (one week, one month, one quarter). The prediction results are presented in the form of trend curve or numerical sequence, clearly showing the possible upward, downward or stable fluctuation trends of future VOCs emission.

[0044] 4.2 Random forest algorithm quantifies the operation efficiency of governance equipment; 4.2.1 Data collection and sorting; Collect the operation-related data of the treatment equipment, including the operation time of the equipment, maintenance records, the amount of VOCs treated, energy consumption, and other operation parameters, as well as the corresponding treatment effect data, such as the change in VOCs concentration before and after treatment, the actual reduction in VOCs emissions, etc. Classify and organize these data to ensure that each set of operation parameters has corresponding treatment effect data to match, forming complete sample data.

[0045] 4.2.2 Model construction and training; Use the random forest algorithm to build the model, and divide the sorted sample data into training samples and test samples. The operation parameters in the training samples are used as the input of the model, and the corresponding treatment effect is used as the learning target of the model. The model analyzes a large number of training samples to find the relationship between different combinations of operation parameters and treatment effects, such as whether too long equipment operation time will lead to a decrease in treatment efficiency, whether there is a reasonable proportional relationship between energy consumption and processing capacity, etc. Test samples are used to evaluate the quantification ability of the model, and the parameters of the model, such as the number of decision trees, are adjusted according to the test results to improve the quantification accuracy of the model on the operation efficiency of the treatment equipment.

[0046] 4.2.3 Operation efficiency quantification; Input the current operation parameters of the treatment equipment into the trained random forest model, and the model will output a quantification value representing the operation efficiency of the treatment equipment according to the learned correlation rules. This quantification value can reflect the performance of the equipment in the current running state, for example, when the quantification value is high, it means that the equipment is in good running state and the treatment effect is good; when the quantification value is low, it may mean that the equipment has running abnormalities and needs to be repaired or adjusted.

[0047] 4.3 Transformer model evaluation of VOCs emission health risk index; 4.3.1 Data collection and processing; Collect data related to the health risks of VOCs emissions, including VOCs concentration distribution data in different areas, population distribution in the area (number of residents, population density, age structure, etc.), activity time of the population in the area (working time, living time, etc.), and toxicity characteristic data of VOCs (harmfulness of different types of VOCs to different organs of the human body, etc.). Integrate and process these data to ensure the accuracy and relevance of the data, for example, match the VOCs concentration in a specific area with the population characteristics of that area.

[0048] 4.3.2 Model training and verification; The processed data is divided into a training part and a validation part. The training part is used to let the Transformer model learn the relationship between VOCS concentration, population exposure, VOCS toxicity, and health risk, such as how high concentration of VOCS in densely populated areas for a long time exposure will bring health risk. Through training, the model can understand the interaction between these complex factors. The validation part is used to test the accuracy of the model in assessing health risks, and according to the validation results, the model is optimized to make the model more suitable for actual health risk assessment needs.

[0049] 4.3.3 Health risk index assessment; Input the VOCS concentration data, population information, and population activity time data of the area to be evaluated into the trained Transformer model. The model will consider various factors and output a health risk index. This index can intuitively reflect the harm degree of VOCS emissions to the health of the population in the area. The higher the index, the greater the health risk.

[0050] 4.4 Establish a dynamic assessment model; 4.4.1 Integration of results of each sub-model; Integrate the future VOCS emission trend predicted by the LSTM neural network, the operation efficiency of the treatment equipment quantified by the random forest algorithm, and the health risk index evaluated by the Transformer model. In the integration process, the role of each result in the dynamic assessment model needs to be clarified, for example, the emission trend reflects the overall trend of future emissions, the operation efficiency of the treatment equipment reflects the current treatment capacity, and the health risk index represents the potential harm caused by emissions.

[0051] 4.4.2 Design of dynamic adjustment mechanism of the model; Design a dynamic adjustment mechanism to enable the model to update the assessment results in real time according to new data input. When new VOCS emission data, treatment equipment operation data, or health impact related data are generated, the model can automatically include these data in the analysis range, recalculate the emission trend, operation efficiency, and health risk index, and adjust the output results of the dynamic assessment model.

[0052] Step 105, input the fusion feature vector into the dynamic assessment model, and output an industrial VOCS emission quality assessment report; Specifically, in this embodiment, the fusion feature vector is input into the dynamic assessment model, and the evaluation result is output. The evaluation result is divided into different levels, including at least excellent, good, qualified, and unqualified.

[0053] Specifically; An overall evaluation result of the quality of industrial VOCs emissions is given, which is based on the evaluation indexes calculated by the dynamic evaluation model. The evaluation result can be divided into different grades, such as excellent, good, qualified, unqualified, etc., and the main basis for obtaining the evaluation result is briefly explained, so that the reader can quickly understand the overall situation of the emission quality.

[0054] 1. Future emission trend and risk warning; According to the future VOCs emission trend predicted by the LSTM neural network, the possible changes in emissions in the future are described in detail in the report, such as whether the emission amount will increase, whether the emission intensity will increase, etc.

[0055] Combined with the future emission trend and the health risk index, the possible emission risk is warned. If it is predicted that the future emission will cause great risk to the surrounding environment and the health of the population, the risk level and the possible time range are clearly pointed out, and the relevant units are reminded to take preventive measures in advance.

[0056] 2. Treatment suggestions; According to the evaluation result, specific treatment suggestions are put forward for the problems existing in the emission quality. If the emission intensity is large, it is suggested to optimize the production process, reduce the consumption of raw materials, etc.; if the treatment efficiency is not high, it is suggested to strengthen the equipment maintenance, adjust the equipment operation parameters or replace the more efficient treatment equipment, etc.

[0057] At the same time, combined with the future emission trend and risk warning, forward-looking prevention and treatment measures are put forward to avoid further deterioration of the emission quality and reduce the emission risk.

[0058] Its beneficial effects are: 1. It realizes the comprehensive characterization of industrial VOCs emissions from multiple dimensions. The emission intensity feature reflects the scale and intensity of emissions, the treatment efficiency feature reflects the effectiveness of treatment measures, and the environmental impact feature shows the effect of emissions on the surrounding environment. This multi-dimensional evaluation method avoids the limitations of traditional single-index evaluation, and the evaluation result can more comprehensively reflect the actual quality of industrial VOCs emissions. 2. It can discover potential risks in time and provide a scientific basis for formulating prevention and response measures, which helps to control the emission risk at a low level.

[0059] Please refer to Figure 2 In a method for evaluating the quality of industrial VOCs emissions based on big data, standard processing multi-source data is transmitted to a cloud platform, and a YOLOv8 target detection model is used in the cloud platform to detect image data in the standard processing multi-source data to obtain image features, including the following steps: Step 201, the calculated unit time emission amount is arranged and analyzed, and the emission intensity feature is obtained by comparing the emission amounts in different time periods and under different production conditions; Step 202, subtract the concentration after treatment from the concentration before treatment, compare the concentration before treatment, and obtain the treatment efficiency characteristics; Step 203, obtain detailed topographic data, including the relief of the terrain, the trend of the mountains and the distribution of rivers, input the topographic data, the emission source data and the meteorological data into the ADMS model; Step 204, simulate the concentration distribution of VOCS at different positions according to the influence of complex terrain on air movement, and obtain the environmental impact characteristics.

[0060] The above describes the embodiments of the industrial VOCS emission quality evaluation method based on big data, please refer to Figure 3 In an industrial VOCS emission quality evaluation system based on big data, the industrial VOCS emission quality evaluation system comprises the following modules: A multi-source data acquisition module is used to acquire multi-source industrial emission data, preprocess the multi-source industrial emission data, and obtain standard industrial emission data; A multi-source feature extraction module is used to obtain emission intensity characteristics based on VOCS concentration and flow rate, calculate treatment efficiency characteristics by comparing VOCS concentration before and after treatment, use AERMOD to quickly predict diffusion range and use ADMS to simulate concentration distribution under complex terrain, and obtain environmental impact characteristics; A feature vector fusion module is used to fuse emission intensity characteristics, treatment efficiency characteristics and environmental impact characteristics through a cross-modal attention mechanism to obtain a fused feature vector; An evaluation model establishment module is used to predict the VOCS emission trend in the future period by using an LSTM neural network, quantify the operation efficiency of the treatment equipment by using a random forest algorithm, evaluate the health risk index of VOCS emission according to a Transformer model, and establish a dynamic evaluation model; An evaluation report generation module is used to input the fused feature vector into the dynamic evaluation model and output an industrial VOCS emission quality evaluation report.

[0061] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above embodiments, and the above embodiments and descriptions in the specification are only preferred examples of the present application and are not intended to limit the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and equivalents.

Claims

1. A big data based industrial VOCS emission quality assessment method, characterized in that, The industrial VOCs emission quality evaluation method comprises the following steps: Obtain multi-source industrial emission data, preprocess the multi-source industrial emission data, and obtain standard industrial emission data; Based on the VOCs concentration and the flow rate, the emission intensity characteristics are calculated, the treatment efficiency characteristics are calculated by comparing the VOCs concentration before and after treatment, the diffusion range is quickly predicted by AERMOD, the concentration distribution under complex terrain is simulated by ADMS, and the environmental impact characteristics are obtained; The emission intensity characteristics, the treatment efficiency characteristics and the environmental impact characteristics are fused through a cross-modal attention mechanism to obtain a fusion feature vector; The future VOCs emission trend is predicted by using an LSTM neural network, the operation efficiency of the treatment equipment is quantified by using a random forest algorithm, the health risk index of VOCs emission is evaluated according to a Transformer model, and a dynamic evaluation model is established; The fusion feature vector is input into the dynamic evaluation model, and an industrial VOCs emission quality evaluation report is output.

2. The big data based industrial VOCS emission quality evaluation method according to claim 1, wherein, The multi-source industrial emission data is obtained, the multi-source industrial emission data is preprocessed, and the standard industrial emission data is obtained, comprising: Obtain emission source data, equipment working condition data, environmental meteorological data and production process data to obtain multi-source industrial emission data; The missing values in the multi-source industrial emission data are processed by using the mean filling method to obtain cleaned industrial emission data; The multi-source data of different magnitudes in the cleaned industrial emission data is converted to the same magnitude by using the Z-score standardization method, and then integrated to obtain the standard industrial emission data.

3. The big data based industrial VOCS emission quality assessment method according to claim 1, wherein, Based on the VOCs concentration and the flow rate, the emission intensity characteristics are calculated, the treatment efficiency characteristics are calculated by comparing the VOCs concentration before and after treatment, the diffusion range is quickly predicted by AERMOD, the concentration distribution under complex terrain is simulated by ADMS, and the environmental impact characteristics are obtained, comprising: The calculated emission amount per unit time is sorted and analyzed, and the emission intensity characteristics are obtained by comparing the emission amounts under different time periods and different production conditions; The concentration before treatment is subtracted from the concentration after treatment, and compared with the concentration before treatment to obtain the treatment efficiency characteristics; Obtain detailed terrain data, including the relief degree of the terrain, the direction of the mountains and the distribution of the rivers, input the terrain data and the emission source data and the meteorological data into the ADMS model; According to the influence of complex terrain on air movement, the concentration distribution of VOCs at different positions is simulated to obtain the environmental impact characteristics.

4. The big data based industrial VOCS emission quality evaluation method according to claim 1, wherein, The emission intensity characteristics, the treatment efficiency characteristics and the environmental impact characteristics are fused through a cross-modal attention mechanism to obtain a fusion feature vector, comprising: The query vector of the emission intensity characteristics is compared with the key vector of the treatment efficiency characteristics, the similarity of the two is calculated, and the attention weight of the emission intensity characteristics to the treatment efficiency characteristics is obtained; The attention weight between the emission intensity characteristics and the environmental impact characteristics, and the attention weight between the treatment efficiency characteristics and the environmental impact characteristics are calculated; According to the calculated attention weight, the value vectors of each feature are weighted and summed, and the value vectors of the emission intensity feature are adjusted according to the corresponding attention weight, and the value vectors of the governance efficiency feature and the environmental impact feature are adjusted to obtain a fusion feature vector.

5. The big data based industrial VOCS emission quality evaluation method according to claim 1, wherein, The dynamic evaluation model is established by using an LSTM neural network to predict the VOCs emission trend in a future period, using a random forest algorithm to quantify the operation efficiency of the governance equipment, and using a Transformer model to evaluate the health risk index of VOCs emission. The data is divided into training samples and test samples by using a random forest algorithm, the operation parameters in the training samples are used as the input of the model, and the corresponding governance effect is used as the learning target of the model. The processed data is divided into a training part and a verification part, and the training part is used to learn the relationship between VOCs concentration, population exposure, VOCs toxicity factors and health risk.

6. The big data based industrial VOCS emission quality evaluation method according to claim 1, wherein, The dynamic evaluation model is established by using an LSTM neural network to predict the VOCs emission trend in a future period, using a random forest algorithm to quantify the operation efficiency of the governance equipment, and using a Transformer model to evaluate the health risk index of VOCs emission. The future VOCs emission trend predicted by the LSTM neural network, the operation efficiency of the governance equipment quantified by the random forest algorithm, and the health risk index evaluated by the Transformer model are integrated to establish a dynamic evaluation model.

7. The big data based industrial VOCS emission quality assessment method of claim 1, wherein, The fusion feature vector is input into the dynamic evaluation model, and an industrial VOCs emission quality evaluation report is output. The fusion feature vector is input into the dynamic evaluation model, and an evaluation result is output, and the evaluation result is divided into different levels, including at least excellent, good, qualified and unqualified.

8. A big data based industrial VOCS emission quality assessment system, characterized in that, The industrial VOCs emission quality evaluation system includes the following modules: A multi-source data acquisition module is used to acquire multi-source industrial emission data, and the multi-source industrial emission data is preprocessed to obtain standard industrial emission data; A multi-source feature extraction module is used to acquire an emission intensity feature based on VOCs concentration and flow rate, calculate a governance efficiency feature by comparing VOCs concentration before and after governance, use AERMOD to quickly predict the diffusion range and use ADMS to simulate the concentration distribution under complex terrain to obtain an environmental impact feature; A feature vector fusion module is used to fuse the emission intensity feature, the governance efficiency feature and the environmental impact feature through a cross-modal attention mechanism to obtain a fusion feature vector; An evaluation model establishment module is used to use an LSTM neural network to predict the VOCs emission trend in a future period, use a random forest algorithm to quantify the operation efficiency of the governance equipment, and use a Transformer model to evaluate the health risk index of VOCs emission to establish a dynamic evaluation model; An evaluation report generation module is used to input the fusion feature vector into the dynamic evaluation model to output an industrial VOCs emission quality evaluation report.

9. The big data based industrial VOCS emission quality evaluation system of claim 8, wherein, The evaluation model establishment module includes the following units: The input unit is configured to divide data into training samples and test samples by using a random forest algorithm, and use operation parameters in the training samples as inputs of a model and use corresponding treatment effects as learning targets of the model. The division unit is configured to divide the processed data into a training part and a verification part, and use the training part to learn relationships among VOCS concentration, population exposure, VOCS toxicity factors and health risks.

10. The big data based industrial VOCS emission quality evaluation system of claim 8, wherein, The evaluation report generation module comprises the following units: The establishment unit is configured to integrate a future VOCS emission trend predicted by an LSTM neural network, an operation efficiency of a treatment device quantified by a random forest algorithm and a health risk index evaluated by a Transformer model, and establish a dynamic evaluation model.

Images