Production enterprise-oriented pollutant degradation scheme recommendation method and system and medium

Abstract

The invention discloses a production enterprise-oriented pollutant degradation scheme recommendation method and system and a medium, and relates to the technical field of environmental protection, and the method comprises the steps: collecting multi-source pollution related data in real time in an enterprise production area, and uploading the data to a data processing center through an Internet of Things transmission network; a pollutant prediction model is constructed, pollutant generation prediction is carried out according to the preprocessed multi-source pollution related data, and the migration and diffusion trend of generated pollutants is analyzed; a pollutant generation prediction result and a pollutant migration and diffusion trend are used to construct a pollution portrait, a degradation technology is selected, and a degradation scheme is generated; and executing the degradation scheme to degrade pollutants, and acquiring operation parameters and degradation effect data of degradation equipment to perform equipment fault monitoring and degradation scheme optimization. According to the invention, through real-time monitoring, accurate data analysis and dynamic adjustment of a degradation scheme, and in combination with a novel composite degradation technology, continuous, efficient and stable treatment of pollutants is realized, and the influence of the pollutants on the environment is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental protection, and more specifically, to a method, system and medium for recommending pollutant degradation solutions for production enterprises. Background Art

[0002] With the increasingly strict environmental protection requirements, production enterprises face huge challenges in pollutant treatment. Traditional pollutant degradation systems have many drawbacks. Shutting down production for equipment maintenance or transformation seriously affects the normal production of enterprises, leading to enterprises' resistance to restoration projects. For example, physical adsorption may cause secondary pollution, chemical oxidation has high energy consumption and may generate harmful intermediate products, while biodegradation is limited by the adaptability and degradation efficiency of microorganisms. Large-scale earthmoving restoration is costly and may cause irreparable damage to the infrastructure within the enterprise site, with low practical operation feasibility. And pollutants are continuously and inevitably generated during the production process, and the existing treatment methods are difficult to meet the requirements.

[0003] Currently, there are many studies on degradation technologies for specific pollutants, but there is a lack of a systematic method that can quickly match the optimal degradation solution according to the actual needs of production enterprises (such as pollutant characteristics, treatment scale, site conditions, cost budget, etc.). In addition, existing technologies often ignore the real-time monitoring and dynamic adjustment capabilities at the production site, resulting in unstable treatment effects or insufficient adaptability. Therefore, how to accurately analyze the generation, migration and diffusion laws of pollutants in order to formulate efficient degradation solutions, ensure the normal operation of enterprises, and effectively treat pollutants is an urgent problem to be solved at present. Summary of the Invention

[0004] To solve the above technical problems, the present invention proposes a method, system and medium for recommending pollutant degradation solutions for production enterprises, aiming to provide production enterprises with efficient and low-disturbance pollutant treatment solutions, thereby improving the efficiency of pollution control, reducing environmental risks, and contributing to sustainable development.

[0005] In the first aspect of the present invention, a method for recommending pollutant degradation solutions for production enterprises is provided, including the following steps: Collect multi-source pollution-related data in real time in the enterprise production area and upload it to the data processing center through the Internet of Things transmission network; Build a pollutant prediction model in the data processing center, predict the generation of pollutants based on the preprocessed multi-source pollution-related data, and analyze the migration and diffusion trends of the generated pollutants; Use the pollutant generation prediction results and the migration and diffusion trends of pollutants to construct a pollution profile, select a degradation technology based on the pollution profile, and generate a degradation solution; Execute the degradation plan to degrade pollutants, and dynamically adjust the parameters of the degradation equipment in real time according to the pollutant generation rate and concentration changes, and obtain the operation parameters of the degradation equipment and the degradation effect data for equipment fault monitoring and degradation plan optimization.

[0006] In this plan, multi-source pollution-related data is collected in real time in the enterprise production area and uploaded to the data processing center through the Internet of Things transmission network. Specifically: Obtain the production process information and production equipment information of the enterprise production area, divide and align the production process information and production equipment information in production links, use different production links for big data retrieval to obtain production pollution examples, and extract production link characteristics and pollution indicators to construct a structured data set; Conduct pollution correlation modeling and training based on the structured data set, perform SHAP analysis on the trained model, calculate the SHAP values of each production link characteristic, represent the degree of influence on pollution, and take the absolute mean value of the SHAP values of each production link characteristic; Sort the production links through the absolute mean value, and select the production links that meet the requirements as key production links according to the preset threshold; Map the SHAP analysis results of the production link characteristics to the production equipment. For each production equipment, summarize the SHAP values of all production link characteristics associated with it, calculate the equipment-level pollution contribution index, and sort according to the pollution contribution index to identify key production equipment; Identify the production key nodes according to the key production links and key production equipment, use multi-modal sensors to collect multi-source pollution-related data of the production key nodes in real time, use blockchain technology to encrypt the collected data, and upload it to the data processing center through the Internet of Things transmission network.

[0007] In this plan, a pollutant prediction model is constructed in the data processing center, and pollutant generation prediction is carried out according to the preprocessed multi-source pollution-related data. Specifically: Construct a pollutant prediction model based on hybrid deep learning, and use the enterprise historical production data and production pollution examples to construct training data and test data to train and verify the pollutant prediction model, and output the trained pollutant prediction model; Perform data preprocessing on the multi-source pollution-related data, use the preprocessed multi-source pollution-related data as the model input, reconstruct the input data into a two-dimensional matrix form, and use the sliding window method to construct a data image; Construct a feature extraction module using the AlexNet network, extract data features from the data images to construct a data feature sequence, import the data feature sequence into a BiLSTM network to capture the forward and backward dependencies of the enterprise production process, introduce an attention mechanism, and use a genetic algorithm to obtain the optimal number of attention heads, time steps, and the number of neurons in the hidden layer; Obtain the hidden state at each time step, use the self-attention mechanism to calculate the weights of different time steps, output the enhanced temporal features, send them to a fully connected layer for pollutant prediction, and obtain the pollutant generation prediction result.

[0008] In this solution, analyze the migration and diffusion trend of the generated pollutants, specifically: Obtain the pollutant category label according to the pollutant generation prediction result, use the pollutant category label to obtain the influencing factors of pollutant migration and diffusion based on the big data retrieval method, construct an initial influencing factor set, and perform feature selection using the mRMR algorithm extended by the Pearson correlation coefficient in the initial influencing factor set; Use the Pearson correlation coefficient to quantify the linear relationship between the influencing factors and pollutant migration and diffusion, select a preset number of influencing factors as the candidate influencing factor set, and calculate the correlation between the influencing factors in the candidate influencing factor set to select the feature with the least redundancy; Import the influencing factors selected by feature selection into an SVM model, generate the importance score of the influencing factors according to the classification accuracy of the model, sort through the importance score, and select the key influencing factors based on the sorting result; Perform grid processing on a preset area range centered on the enterprise production area, use the grid blocks as nodes, construct node feature vectors according to the key influencing factors and pollutant monitoring concentrations, establish an edge structure according to the actual distance between nodes and the correlation between node feature vectors, and construct a spatio-temporal graph; Learn the spatio-temporal graph through a graph convolutional network, perform message passing on the node feature vectors according to graph convolution, and introduce a spatial attention mechanism to aggregate neighboring nodes to extract spatial features; Extract the time dependence of the node feature vectors through a gated temporal unit, introduce self-attention to weight the features, extract time features, fuse the spatial features and the time features to generate spatio-temporal features, and import the spatio-temporal features into a prediction head to generate the pollutant concentration prediction result of the grid block; Interpolate based on the pollutant concentration prediction result to generate a continuous field, calculate the gradient direction of the pollutant concentration according to the continuous field, and generate the migration and diffusion trend of the generated pollutants.

[0009] In this solution, use the pollutant generation prediction result and the migration and diffusion trend of the pollutants to construct a pollution portrait, and select a degradation technology based on the pollution portrait to generate a degradation plan, specifically: Generate a pollution source portrait based on the pollutant generation prediction results, generate a migration and diffusion portrait based on the migration and diffusion trend of pollutants, and construct a pollution portrait by combining the pollution source portrait and the migration and diffusion portrait with the degradation difficulty index; Obtain degradation technologies, construct nodes with degradation technologies and pollutant types, establish an edge structure based on the association between degradation technologies and pollutants and the association between degradation technologies, and construct a degradation technology association network through the nodes and the edge structure; Use the pollution portrait to perform matching in the degradation technology association network, obtain the initial degradation technology nodes, obtain the usage frequency and treatment efficiency of the initial degradation technology nodes according to pollutant degradation instances, and obtain the interested degradation technologies corresponding to the pollution portrait; Use random walk to start from the initial degradation technology nodes, generate a walk path through iterative walking, record the technical combination effects on the path, calculate the multi-objective scores of the path according to the average treatment efficiency, cost index and technology fitness, and select a preset number of walk paths; Stitch the interested degradation technologies in the selected set of walk paths, obtain the encoding of each stitched walk path to generate a path matrix, use the attention mechanism to generate a weight vector for the path matrix, select the degradation technology combination corresponding to the walk path with the maximum weight vector, and generate a degradation plan.

[0010] In this solution, execute the degradation plan to degrade pollutants, and dynamically adjust the parameters of the degradation equipment in real time according to the pollutant generation rate and concentration changes. Specifically: Obtain the corresponding degradation equipment and equipment parameters according to the degradation plan, generate a modular deployment plan based on the degradation equipment and equipment parameters according to the pollution portrait, and execute pollutant degradation; Monitor the pollutant generation rate and concentration changes. When the pollutant generation rate and concentration changes are greater than the preset threshold, initiate parameter optimization, set the adjustment intensity according to the concentration deviation range where the deviation value falls, and dynamically adjust the parameters of the degradation equipment according to the adjustment intensity.

[0011] In this solution, obtain the operation parameters of the degradation equipment and the degradation effect data for equipment fault monitoring and degradation plan optimization. Specifically: Real-time monitor the operation status of the key components of the degradation equipment according to multiple sensors of the degradation equipment, train a data-driven model in the data processing center according to the normal operation status data, use the data-driven model to identify the abnormal operation status of the degradation equipment, generate a fault warning and automatically adjust the equipment operation parameters; Collect samples at the monitoring points to analyze the pollutant concentration, feedback the monitoring data to the data processing center, compare it with the preset degradation target. If it does not meet the expectation, optimize the degradation plan to form a closed-loop optimization feedback mechanism.

[0012] The second aspect of the present invention provides a pollutant degradation solution recommendation system for production enterprises, realizing a method for recommending pollutant degradation solutions for production enterprises. The system includes a data acquisition unit, a data processing central unit, a degradation solution formulation and adjustment unit, a degradation equipment operation monitoring unit, and a degradation effect monitoring and feedback unit; The data acquisition unit uses sensors to collect multi-source pollution-related data in real time in the enterprise production area and uploads it to the data processing central unit through the Internet of Things transmission network; The data processing central unit constructs a pollutant prediction model using deep learning and conducts model training, predicts pollutant generation based on the pre-processed multi-source pollution-related data, and analyzes the migration and diffusion trends of the generated pollutants; The degradation solution formulation and adjustment unit uses the pollutant generation prediction results and the migration and diffusion trends of pollutants to construct a pollution profile, selects degradation technologies based on the pollution profile, generates a degradation solution, and modularly deploys pollutant degradation equipment according to the degradation solution; The degradation equipment operation monitoring unit sets the equipment execution parameters according to the degradation solution through an automated control system, real-time feedbacks the operation status for fault monitoring, and dynamically adjusts the degradation equipment parameters in real time according to the changes in the pollutant generation rate and concentration; The degradation effect monitoring and feedback unit collects pollutant residual concentration to analyze the degradation effect data, compares the degradation effect data with the preset degradation target. If the expected level is not reached, the degradation solution is optimized through feedback.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention uses the modular and miniaturized design of in-situ degradation equipment and the operation mode of the system to minimize the interference with the normal production and operation activities of enterprises and ensure the continuous operation of enterprises. The intelligent fault diagnosis and self-repair functions of the equipment further reduce the risk of production interruption caused by equipment failures.

[0014] Through real-time monitoring, accurate data analysis, and dynamic adjustment of the degradation solution, combined with new composite degradation technologies, continuous, efficient, and stable treatment of pollutants is achieved, effectively reducing the impact of pollutants on the environment. Avoid high-cost operations such as large-scale earthwork restoration, reduce the pollutant treatment cost of enterprises, and at the same time reduce the economic losses caused by production suspension. The intelligent maintenance function of the equipment reduces the manual maintenance cost and equipment repair cost.

[0015] The effect monitoring and feedback optimization mechanism ensures that the degradation effect always moves towards the preset target, continuously improving the quality and efficiency of pollutant treatment. The blockchain encryption of data and the pollutant traceability function enhance the management value of the system and also help reduce potential compliance costs in the long run. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] To more clearly illustrate the technical solutions in the embodiments or exemplifications of the present invention, the following will briefly introduce the drawings required for use in the embodiments or exemplifications. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings shown.

[0017] Figure 1 Shows the flowchart of the pollutant degradation solution recommendation method for production enterprises; Figure 2 Shows the flowchart of the embodiment for constructing a pollutant prediction model to predict pollutant generation; Figure 3 Shows the flowchart of the embodiment for generating a degradation solution based on the pollution portrait; Figure 4 Shows the block diagram of the pollutant degradation solution recommendation system for production enterprises. Detailed implementation manners

[0018] In order to more clearly understand the above objects, features, and advantages of the present invention, the present invention will be further described in detail below in conjunction with the drawings and specific implementation manners. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0019] Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0020] Figure 1 Shows the flowchart of the pollutant degradation solution recommendation method for production enterprises.

[0021] As Figure 1 shown, this embodiment provides a pollutant degradation solution recommendation method for production enterprises, including: S102, collecting multi-source pollution-related data in real time in the enterprise production area and uploading it to the data processing center through the Internet of Things transmission network; S104, constructing a pollutant prediction model in the data processing center, predicting pollutant generation according to the preprocessed multi-source pollution-related data, and analyzing the migration and diffusion trend of the generated pollutants; S106, using the pollutant generation prediction result and the migration and diffusion trend of the pollutants to construct a pollution portrait, selecting a degradation technology based on the pollution portrait, and generating a degradation solution; In S108, execute the degradation scheme to degrade pollutants, and dynamically adjust the parameters of the degradation equipment in real time according to the pollutant generation rate and concentration changes, obtain the operation parameters of the degradation equipment and the degradation effect data for equipment fault monitoring and degradation scheme optimization.

[0022] It should be noted that in the enterprise production area and the pollutant generation source, such as key nodes of production equipment, raw material storage areas, etc., high-precision sensors are deployed, including gas sensors, water quality sensors, soil sensors, etc. The sensors collect information such as pollutant types, concentrations, and generation rates in real time, and upload it to the data processing center through the Internet of Things transmission network. Through the enterprise production management system (MES), equipment monitoring system (SCADA) or manual investigation, obtain detailed process flow information of the production area, including input / output materials, energy consumption, equipment operation parameters, etc. in each production link; collect basic information of production equipment, such as equipment type, operation status, maintenance records, energy consumption data, etc., divide and align the production process information and production equipment information by production link, and establish the association relationship between each link and the corresponding pollutants. Use big data retrieval in different production links to obtain production pollution examples, extract production link characteristics such as equipment operation duration and raw material usage, and pollution indicators such as emission concentration, total amount, and exceeding standard conditions to construct a structured data set.

[0023] Conduct pollution correlation modeling and training based on the structured data set, perform SHAP analysis on the trained model, and quantify the importance of each production link characteristic to pollution contribution. Calculate the SHAP values of each production link characteristic, which represent the degree of influence on pollution (a positive value indicates an increase in pollution, and a negative value indicates a decrease in pollution), take the absolute mean of the SHAP values of each production link characteristic to measure the overall contribution of this production link to pollution; sort the production links through the absolute mean, select the production links that meet the requirements according to the preset threshold, and identify the key production links with higher pollution contribution; map the SHAP analysis results of production link characteristics to production equipment. For each production equipment, summarize the SHAP values of all associated production link characteristics, calculate the equipment-level pollution contribution index, sort according to the pollution contribution index, and identify key production equipment; identify production key nodes according to the key production links and key production equipment, use multi-modal sensors to collect multi-source pollution-related data of production key nodes in real time, and use blockchain technology to encrypt the collected data to ensure the security and non-tampering of the data during transmission and storage, ensure the authenticity and reliability of the data, enhance the trust of enterprises and regulatory authorities in the data, and upload it to the data processing center through the Internet of Things transmission network.

[0024] Figure 2 The flowchart showing the construction of a pollutant prediction model for pollutant generation prediction in an embodiment is shown.

[0025] According to an embodiment of the present invention, a pollutant prediction model is constructed in the data processing center, and pollutant generation prediction is performed according to the preprocessed multi-source pollution-related data. Specifically: S202, construct a pollutant prediction model based on hybrid deep learning, and train and verify the pollutant prediction model according to the enterprise historical production data and production pollution examples to construct training data and test data, and output the trained pollutant prediction model; S204, perform data preprocessing on the multi-source pollution-related data, use the preprocessed multi-source pollution-related data as model input, reconstruct the input data into a two-dimensional matrix form, and use the sliding window method to construct a data image; S206, use the AlexNet network to construct a feature extraction module, extract data features from the data image to construct a data feature sequence, import the data feature sequence into the BiLSTM network to capture the forward and backward dependencies of the enterprise production process, introduce an attention mechanism, and use a genetic algorithm to obtain the optimal number of attention heads, time steps, and the number of hidden layer neurons; S208, obtain the hidden state at each time step, calculate the weights of different time steps using the self-attention mechanism, output the enhanced temporal features, send them to the fully connected layer for pollutant prediction, and obtain the pollutant generation prediction result.

[0026] It should be noted that data preprocessing of the multi-source pollution-related data includes spatio-temporal alignment, outlier removal, and data standardization, etc. The AlexNet network has 5 convolutional layers and 3 fully connected layers. The multi-layer convolutional structure can extract features of different granularities at the same time and can better capture the non-linear features in the production data. Compared with traditional neural network methods, the AlexNet network uses the ReLU activation function to improve the problems of gradient disappearance and convergence fluctuations. At the same time, the DropOut method is introduced to control overfitting. The genetic algorithm is used for efficient search to avoid the time-consuming problem of manual network parameter tuning, and 50 groups of hyperparameters (chromosomes) are randomly generated. The model is trained for each group of parameters, the RMSE of the validation set is calculated and converted into fitness, high-fitness individuals are selected, crossed to generate new individuals, mutations are introduced to introduce diversity, and the hyperparameter combination with the highest fitness in the final population is selected to deploy the AlexNet-BiLSTM-Attention model. Through the automatic optimization of hyperparameters, the pollutant prediction accuracy is significantly improved.

[0027] The bidirectional LSTM (BiLSTM) is adopted to capture the forward and backward dependencies of the production process (such as the impact of upstream processes on downstream pollution), obtain the hidden state at each time step for subsequent Attention calculation, and the self-attention mechanism is used to calculate the weights at different time steps, identify the key pollution stages, obtain the enhanced temporal features, and send them to the fully connected layer to predict the pollutant output using the Sigmoid activation.

[0028] It should be noted that the migration and diffusion trend of the generated pollutants is analyzed. According to the pollutant generation prediction result, the pollutant category label is obtained, and the influencing factors of pollutant migration and diffusion are obtained based on the big data retrieval method using the pollutant category label, and an initial influencing factor set is constructed, including pollution source characteristic factors (emission intensity, emission mode, physical state and chemical properties of pollutants, etc.), environmental medium factors (atmospheric environment, water environment, soil environment, etc.), geographical characteristic factors (topographic elevation data, surface roughness, water body distribution, etc.), and human intervention factors (emergency treatment measures, traffic flow, etc.). The mRMR algorithm extended by the Pearson correlation coefficient is used for feature selection in the initial influencing factor set; the Pearson correlation coefficient is used to quantify the linear relationship between the influencing factors and pollutant migration and diffusion, and a preset number of influencing factors are selected as the candidate influencing factor set, and the correlation between the influencing factors is calculated in the candidate influencing factor set to select the features with the least redundancy; the mRMR score affecting the timbre is obtained with the goal of maximum correlation and minimum redundancy, and the influencing factors of feature selection are obtained after reaching the termination condition. The influencing factors of feature selection are imported into the SVM model, the importance score of the influencing factors is generated according to the classification accuracy of the model, sorted by the importance score, and the key influencing factors are selected based on the sorting result.

[0029] Perform grid processing on a preset area range centered on the enterprise production area, use grid blocks as nodes, construct node feature vectors based on key influencing factors and pollutant monitoring concentrations, establish edge structures according to the actual distances between nodes and the correlations between node feature vectors, and construct a spatio-temporal graph; learn the spatio-temporal graph through a graph convolutional network, perform message passing on the node feature vectors according to graph convolution, and introduce a spatial attention mechanism to aggregate neighborhood nodes to extract spatial features; extract the time dependencies of the node feature vectors through a gated temporal unit, and introduce self-attention to weight the features to extract time features, fuse the spatial features and the time features to generate spatio-temporal features, import the spatio-temporal features into a prediction head to generate a pollutant concentration prediction result for the grid block; perform interpolation based on the pollutant concentration prediction result to generate a continuous field, preferably using radial basis function interpolation, calculate the gradient direction of the pollutant concentration according to the continuous field, and generate the migration and diffusion trend of the generated pollutants. The spatio-temporal graph neural network effectively integrates monitoring network data and physical diffusion mechanisms, improves calculation efficiency and prediction accuracy, and provides intelligent decision-making support for environmental pollution emergency response.

[0030] Figure 3 The flowchart showing the degradation scheme generation based on the pollution portrait in the embodiment is shown.

[0031] According to an embodiment of the present invention, a pollution portrait is constructed using the pollutant generation prediction result and the migration and diffusion trend of the pollutant, and a degradation technology is selected based on the pollution portrait to generate a degradation scheme. Specifically: S302, generate a pollution source portrait according to the pollutant generation prediction result, generate a migration and diffusion portrait according to the migration and diffusion trend of the pollutant, and construct a pollution portrait by combining the pollution source portrait and the migration and diffusion portrait with the degradation difficulty index; S304, obtain degradation technologies, construct nodes with the degradation technologies and pollutant types, establish edge structures according to the associations between the degradation technologies and pollutants and the associations between the degradation technologies, and construct a degradation technology association network through the nodes and edge structures; S306, use the pollution portrait to match in the degradation technology association network to obtain initial degradation technology nodes, obtain the usage frequency and treatment efficiency of the initial degradation technology nodes according to pollutant degradation instances, and obtain the interested degradation technologies corresponding to the pollution portrait; S308, start from the initial degradation technology nodes using random walk, generate a walk path through iterative walks, record the combined effects of the technologies on the path, calculate the multi-objective scores of the paths according to the average treatment efficiency, cost index, and technology fitness, and select a preset number of walk paths; In S310, splice the degradation technology of interest in the selected set of random walk paths, obtain the coding of each spliced random walk path to generate a path matrix, use the attention mechanism to generate a weight vector for the path matrix, select the degradation technology combination corresponding to the random walk path with the maximum weight vector, and generate a degradation plan.

[0032] It should be noted that the pollution source portrait includes the pollutant generation rate, pollutant categories and proportions, emission spatio-temporal characteristics, etc., and the migration and diffusion portrait includes the diffusion influence range, concentration dynamic change trend, sensitive areas, etc. The degradation difficulty index is calculated by the formula: , is the maximum value of pollutant concentration, is the environmental standard limit value, is the pollutant half-life, is the reference value of the pollutant half-life period. Use big data retrieval to obtain degradation technologies, such as activated carbon adsorption, photocatalytic oxidation, etc., perform feature coding on the characteristics of the degradation technology, such as technology type, treatment efficiency, cost index, and applicable pollutants, to construct a degradation technology library.

[0033] In the random walk of the technology association network, calculate the multi-objective score of the path according to the average treatment efficiency, cost index, and technology fitness. Obtain the average treatment efficiency of the degradation technology through normalization calculation, obtain the cost index of the degradation technology through 0-normalization, obtain the technology fitness according to the cosine similarity with the pollution portrait, and construct a multi-objective optimization function by weighted summation of the three objectives. Starting from the initial technology node matched by the pollution portrait, select the next node at each step according to the transition probability, record the technical combination effect on the path, calculate the multi-objective score for each complete path, and retain the Top-5 candidate solutions. Based on the BiLSTM network, obtain the coding of each random walk path, generate a path matrix according to the bidirectional hidden layer features, use the attention mechanism to generate a weight vector for the path matrix, and select the best degradation technology combination. Select appropriate degradation technologies for different pollutants. In addition to traditional organic pollutants using biodegradation and photocatalytic degradation, and heavy metal pollutants using chemical precipitation, ion exchange, etc., select the best degradation technology combination for the random walk path, combine multiple degradation methods, such as combining biodegradation and electrochemical degradation, to exert the advantages of multiple technologies in the same reaction system and improve the degradation effect on mixed pollutants. In addition, the random walk is visually displayed to show the decision-making logic, improve the interpretability of the degradation plan, respond to the changes in the migration and diffusion trend in real time, and achieve a rapid response from pollution prediction to technology recommendation.

[0034] It should be noted that modular design enables the equipment to be flexibly combined according to actual treatment requirements and is applicable to the remediation of contaminated sites of different scales. Obtain the corresponding degradation equipment and equipment parameters according to the degradation plan, and generate a modular deployment plan based on the degradation equipment and equipment parameters according to the pollution profile. The modular deployment plan includes the optimal and backup degradation technical routes to perform pollutant degradation. For example, in-situ soil remediation equipment injects degradation agents by means of injection wells, extraction wells, etc. Miniaturized equipment is more suitable for dispersed pollution sources or sites with limited space, which is convenient for installation and use. For example, vehicle-mounted in-situ injection equipment can achieve precise injection of repair agents and is applicable to complex contaminated sites. Through the intelligent control system and wireless transmission technology, real-time monitoring and optimization of the repair process are realized, improving the repair efficiency and economy. Monitor the pollutant generation rate and concentration changes. When the pollutant generation rate and concentration changes are greater than the preset threshold, trigger parameter optimization. Set the adjustment intensity according to the concentration deviation range where the deviation value falls. Different adjustment intensities correspond to different response times, and dynamically adjust the degradation equipment parameters according to the adjustment intensity.

[0035] Equip the degradation equipment with intelligent fault diagnosis and self-repair functions. Through a variety of built-in sensors and intelligent algorithms, real-time monitor the operating status of its key components, predict potential faults, and automatically perform self-repair or adjust operating parameters at the initial stage of the fault, reducing manual intervention and equipment downtime, and improving the operating efficiency and stability of the equipment. Real-time monitor the operating status of the key components of the degradation equipment according to a variety of sensors. Preferably, in the data processing center, train a data-driven model based on the normal operating status data. For example, use methods such as autoencoder networks, multi-source state estimation, and novelty monitoring models to model the normal behavior of the equipment, construct a data-driven model corresponding to the normal operating status data of the degradation equipment, use the data-driven model to identify the abnormal operating status of the degradation equipment, generate a fault warning, and automatically adjust the equipment operating parameters. Build a fault emergency strategy library. When an abnormal operating status is detected, compare the abnormal characteristics of the abnormal operating status with the triggering conditions of different levels of emergency responses for similarity, and select appropriate countermeasures. For example, when a single device fails, start the standby unit. When the pollutant concentration continues to exceed the standard, switch to the backup degradation technical route.

[0036] Set monitoring points in the surrounding environment of the enterprise, regularly collect samples at the monitoring points to analyze the pollutant concentration, and feedback the monitoring data to the data processing center. Compare it with the preset degradation target. If it does not meet the expectation, optimize the degradation plan to form a closed-loop optimization feedback mechanism.

[0037] Figure 4 The block diagram of the pollutant degradation plan recommendation system for production enterprises is shown.

[0038] The second embodiment of the present invention provides a pollutant degradation solution recommendation system 4 for production enterprises. The system includes a data acquisition unit 401, a data processing center unit 402, a degradation solution formulation and adjustment unit 403, a degradation equipment operation monitoring unit 404, and a degradation effect monitoring and feedback unit 405; The data acquisition unit uses sensors to collect multi-source pollution-related data in real time in the enterprise production area and uploads it to the data processing center unit through the Internet of Things transmission network; The data processing center unit uses deep learning to construct a pollutant prediction model and perform model training, predicts the generation of pollutants based on the preprocessed multi-source pollution-related data, and analyzes the migration and diffusion trends of the generated pollutants; The degradation solution formulation and adjustment unit uses the pollutant generation prediction results and the migration and diffusion trends of pollutants to construct a pollution profile, selects a degradation technology based on the pollution profile, generates a degradation solution, and modularly deploys pollutant degradation equipment according to the degradation solution; The degradation equipment operation monitoring unit sets the equipment execution parameters according to the degradation solution through an automated control system, real-time feedbacks the operation status for fault monitoring, and dynamically adjusts the degradation equipment parameters in real time according to the changes in the pollutant generation rate and concentration; The degradation effect monitoring and feedback unit collects pollutant residual concentration to analyze the degradation effect data, compares the degradation effect data with the preset degradation target, and if the expectation is not met, optimizes the degradation solution through feedback.

[0039] The third embodiment of the present invention provides a computer-readable storage medium, which includes a program for the method of recommending a pollutant degradation solution for production enterprises. When the program for the method of recommending a pollutant degradation solution for production enterprises is executed by a processor, it realizes the steps of the method of recommending a pollutant degradation solution for production enterprises.

[0040] In several embodiments provided by the present application, it should be understood that the disclosed method can be implemented in other ways. The above-described embodiments are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces. The indirect coupling or communication connection of the units can be electrical, mechanical, or other forms. Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps including the above method embodiments; and the foregoing storage medium includes: mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks and other various media that can store program codes.

[0041] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

Claims

1. A method for recommending pollutant degradation plans for production enterprises, characterized in that: The following steps are involved: Collect multi-source pollution-related data in real time in the enterprise production area and upload it to the data processing center through the Internet of Things transmission network; Constructing a pollutant prediction model in the data processing center, predicting pollutant generation based on pre-processed multi-source pollution-related data, and analyzing the migration and diffusion trends of generated pollutants; Construct a pollution profile using the pollutant generation prediction results and the migration and diffusion trends of the pollutants, select a degradation technology based on the pollution profile, and generate a degradation plan; The degradation scheme is executed to degrade pollutants, and according to the pollutant generation rate and concentration changes, the degradation equipment parameters are adjusted dynamically in real time, and the degradation equipment operating parameters and degradation effect data are obtained to monitor equipment failures and optimize the degradation scheme.

2. The method for recommending pollutant degradation solutions for production enterprises according to claim 1, characterized in that: Collect multi-source pollution-related data in real time in the production area of ​​the enterprise and upload it to the data processing center through the Internet of Things transmission network. Specifically: Obtain the production process information and production equipment information of the enterprise's production area, divide and align the production process information and production equipment information into production links, use different production links to perform big data retrieval to obtain production pollution instances, extract production link characteristics and pollution indicators to construct a structured data set; Conduct pollution correlation modeling and training according to the structured data set, conduct SHAP analysis on the trained model, calculate the SHAP value of each production link feature, characterize the degree of impact on pollution, and take the absolute mean of the SHAP value of each production link feature; The production links are sorted by the absolute mean, and the production links that meet the requirements are selected as key production links according to a preset threshold; Map the SHAP analysis results of the production link characteristics to the production equipment. For each production equipment, summarize the SHAP values ​​of all the production link characteristics associated with it, calculate the equipment-level pollution contribution index, sort according to the pollution contribution index, and identify the key production equipment; According to the key production links and key production equipment, the key production nodes are identified, and multi-source pollution-related data of the key production nodes are collected in real time using multimodal sensors. The collected data is encrypted using blockchain technology and uploaded to the data processing center through the Internet of Things transmission network.

3. The method for recommending pollutant degradation solutions for production enterprises according to claim 1, characterized in that: A pollutant prediction model is constructed in the data processing center to predict pollutant generation based on the pre-processed multi-source pollution-related data, specifically: Construct a pollutant prediction model based on hybrid deep learning, construct training data and test data based on the company's historical production data and production pollution examples to train and verify the pollutant prediction model, and output the trained pollutant prediction model; Preprocessing the multi-source pollution related data, using the preprocessed multi-source pollution related data as model input, reconstructing the input data into a two-dimensional matrix form, and constructing a data image using a sliding window method; A feature extraction module is constructed using the AlexNet network, data features are extracted from the data image to construct a data feature sequence, the data feature sequence is imported into the BiLSTM network to capture the front-end dependency of the enterprise production process, an attention mechanism is introduced, and a genetic algorithm is used to obtain the optimal number of attention heads, time steps, and number of hidden layer neurons; Obtain the hidden state of each time step, use the self-attention mechanism to calculate the weights of different time steps, output the enhanced time series features, send them to the fully connected layer for pollutant prediction, and obtain the pollutant generation prediction results.

4. The method for recommending pollutant degradation solutions for production enterprises according to claim 1, characterized in that: Analyze the migration and diffusion trends of the pollutants that have been generated, specifically: Obtain pollutant category labels based on pollutant generation prediction results, use the pollutant category labels to obtain influencing factors of pollutant migration and diffusion based on a big data retrieval method, construct an initial influencing factor set, and use the mRMR algorithm expanded by the Pearson correlation coefficient to perform feature selection in the initial influencing factor set; The linear relationship between the influencing factors and the migration and diffusion of pollutants is quantified using the Pearson correlation coefficient, a preset number of influencing factors are selected as a candidate influencing factor set, and the correlation between the influencing factors is calculated in the candidate influencing factor set to select the feature with the least redundancy; Obtain the influencing factors of feature selection and import them into the SVM model, generate importance scores of the influencing factors according to the classification accuracy of the model, sort them according to the importance scores, and select key influencing factors based on the sorting results; The preset area centered on the enterprise production area is processed by gridding, and the grid blocks are used as nodes. The node feature vectors are constructed according to the key influencing factors and the pollutant monitoring concentration. The edge structure is established according to the actual distance between nodes and the correlation between the node feature vectors to construct a space-time graph. The spatiotemporal graph is learned through a graph convolutional network, node feature vectors are message-transmitted according to graph convolution, and a spatial attention mechanism is introduced to aggregate neighboring nodes to extract spatial features; The time dependency of the node feature vector is extracted by the gated timing unit, and self-attention is introduced to weight the features to extract the time features, the spatial features and the time features are fused to generate the spatiotemporal features, and the spatiotemporal features are imported into the prediction head to generate the pollutant concentration prediction results of the grid block; Based on the pollutant concentration prediction result, interpolation is performed to generate a continuous field, and the gradient direction of the pollutant concentration is calculated according to the continuous field to generate the migration and diffusion trend of the generated pollutants.

5. The method for recommending pollutant degradation solutions for production enterprises according to claim 1, characterized in that: The pollution profile is constructed using the pollutant generation prediction results and the migration and diffusion trends of pollutants. Based on the pollution profile, the degradation technology is selected and the degradation plan is generated, specifically: Generate a pollution source portrait based on the pollutant generation prediction results, generate a migration and diffusion portrait based on the migration and diffusion trend of the pollutants, and construct a pollution portrait by combining the pollution source portrait and the migration and diffusion portrait with the degradation difficulty index; Obtain degradation technology, construct nodes for degradation technology and pollutant types, establish edge structures based on the association between degradation technology and pollutants and between degradation technologies, and construct a degradation technology association network through nodes and edge structures; Use the pollution profile to match in the degradation technology association network to obtain the initial degradation technology node, obtain the usage frequency and processing efficiency of the initial degradation technology node according to the pollutant degradation instance, and obtain the degradation technology of interest corresponding to the pollution profile; Use random walk to start from the initial degradation technology node, generate a walk path through iterative walk, record the technology combination effect on the path, calculate the multi-objective score of the path based on the average processing efficiency, cost index and technology fitness, and select a preset number of walk paths; The degradation technologies of interest are spliced ​​in the selected wandering paths, the encoding of each wandering path after splicing is obtained to generate a path matrix, a weight vector is generated for the path matrix using the attention mechanism, and the degradation technology combination corresponding to the wandering path with the largest weight vector is selected to generate a degradation plan.

6. The method for recommending pollutant degradation solutions for production enterprises according to claim 1, characterized in that: The degradation scheme is implemented to degrade pollutants, and the degradation equipment parameters are dynamically adjusted in real time according to the pollutant generation rate and concentration changes, specifically: Acquire corresponding degradation equipment and equipment parameters according to the degradation plan, generate a modular deployment plan based on the degradation equipment and equipment parameters according to the pollution profile, and perform pollutant degradation; Monitor the pollutant generation rate and concentration changes, trigger parameter optimization when the pollutant generation rate and concentration changes are greater than the preset threshold, set the adjustment intensity according to the concentration deviation range within which the deviation value falls, and dynamically adjust the degradation equipment parameters according to the adjustment intensity.

7. The method for recommending pollutant degradation solutions for production enterprises according to claim 1, characterized in that: Obtain the operating parameters and degradation effect data of the degradation equipment to monitor equipment failures and optimize the degradation plan, specifically: The operating status of the key components of the degradation equipment is monitored in real time by various sensors of the degradation equipment, and a data-driven model is trained in the data processing center according to the normal operating status data, and the data-driven model is used to identify the abnormal operating status of the degradation equipment, generate fault warnings and automatically adjust the equipment operating parameters; Collect samples from monitoring points to analyze pollutant concentrations, feed the monitoring data back to the data processing center, and compare it with the preset degradation target. If it does not meet expectations, optimize the degradation plan to form a closed-loop optimization feedback mechanism.

8. A pollutant degradation solution recommendation system for production enterprises, characterized in that: Implementing the pollutant degradation plan recommendation method for production enterprises as described in any one of claims 1 to 7, the system includes a data acquisition unit, a data processing center unit, a degradation plan formulation and adjustment unit, a degradation equipment operation monitoring unit and a degradation effect monitoring feedback unit; The data collection unit uses sensors to collect multi-source pollution-related data in real time in the enterprise production area, and uploads it to the data processing central unit through the Internet of Things transmission network; The data processing central unit uses deep learning to construct a pollutant prediction model and conducts model training, predicts pollutant generation based on pre-processed multi-source pollution-related data, and analyzes the migration and diffusion trends of generated pollutants; The degradation plan formulation and adjustment unit uses the pollutant generation prediction results and the migration and diffusion trends of the pollutants to construct a pollution portrait, selects degradation technology based on the pollution portrait, generates a degradation plan, and modularly deploys pollutant degradation equipment according to the degradation plan; The degradation equipment operation monitoring unit sets equipment execution parameters according to the degradation plan through an automated control system, provides real-time feedback of the operation status for fault monitoring, and dynamically adjusts degradation equipment parameters in real time according to pollutant generation rate and concentration changes; The degradation effect monitoring feedback unit collects pollutant residual concentration analysis degradation effect data, compares the degradation effect data with the preset degradation target, and optimizes the degradation plan through feedback if the expectation is not met.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a pollutant degradation scheme recommendation method program for production enterprises. When the pollutant degradation scheme recommendation method program for production enterprises is executed by a processor, the steps of the pollutant degradation scheme recommendation method for production enterprises as described in any one of claims 1 to 7 are implemented.

Images