A safe and environmentally-friendly management system and method based on multi-pollutant collaborative treatment

Abstract

The application discloses a kind of based on the safe environmental protection management system and method of multi-pollutant collaborative management, it is related to environmental management technical field;The present application acquires multi-pollution data in real time by multi-parameter sensor network;Optimized pollutant correlation model is constructed using graph neural network and Nelder-Mead simplex method optimization, the collaborative / antagonistic relationship between pollutants is analyzed, and multi-pollutant treatment list is generated;Through improved genetic algorithm, three-dimensional objective function including removal rate, cost and byproduct amount is optimized, and the optimal solution is selected, the balance of environmental and economic benefits is realized, dynamic multi-pollution atlas is constructed, and pollution migration path is accurately tracked;Establish byproduct material flow network analysis system, realize resource recycling through directional separation and reprocessing technology;The present application improves the resolution accuracy of complex pollution, reduces the treatment cost, and reduces the incidence of secondary pollution.

Description

Technical Field

[0001] This invention belongs to the field of environmental management technology, and specifically relates to a safety and environmental protection management system and method based on the collaborative treatment of multiple pollutants. Background Technology

[0002] Chinese patent CN118521165B discloses a pollutant monitoring method and system for a smart factory. The invention covers four steps: acquiring pollutant information, analyzing pollutant hazards, analyzing and handling hazard distances, and treating pollutants. The invention monitors the dust pollutant concentration of each manufacturing machine cluster at different directions and distances, and promptly identifies and addresses the problem of excessive dust levels.

[0003] Existing technologies employ a single-pollutant monitoring and step-by-step treatment model, which is insufficient to address the complex synergistic effects between pollutants. Optimization targets are often singular, with removal rate as the sole guiding principle, neglecting the trade-offs between treatment costs, equipment energy consumption, and secondary pollution from byproducts. Byproduct treatment methods are crude, mainly relying on landfill or direct discharge, resulting in low resource recovery rates and a high risk of secondary environmental burden. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] To address the problems in related technologies, this invention provides a safety and environmental management method based on the synergistic treatment of multiple pollutants, thereby overcoming the aforementioned technical problems existing in the existing related technologies.

[0006] (II) Technical Solution

[0007] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0008] S1. Detect multiple pollutants to obtain real-time multi-pollutant data;

[0009] The pollutants corresponding to the real-time pollutant data that are dangerous are directly processed so that all real-time pollutants are no longer dangerous, resulting in dangerous real-time multi-pollutant data.

[0010] S2. Input the real-time multi-pollutant data that poses no danger into the optimized pollutant correlation model obtained by combining historical data with optimization algorithms to obtain real-time multi-pollutant relationship data;

[0011] S3. Construct a real-time multi-pollutant treatment list based on real-time multi-pollutant relationship data;

[0012] S4. Based on the constraints and the real-time multi-pollutant treatment list, an initial set of treatment schemes is obtained;

[0013] The optimal governance scheme is obtained by combining the initial governance scheme set with an optimization algorithm;

[0014] S5. Based on the optimal solution and dynamic pollution map, conduct synergistic treatment of multiple pollutants;

[0015] This invention integrates multi-source data fusion, deep learning algorithm collaborative optimization, and closed-loop safety management technologies to construct a dynamic pollution map and an optimized pollutant correlation model. This effectively addresses the limitations of traditional pollutant treatment methods, such as singular treatment approaches, removal rate-oriented limitations, and improper byproduct disposal. Based on a multi-parameter sensor network, the system integrates historical data and intelligent optimization algorithms to construct a pollutant correlation model, accurately analyzing the synergistic / antagonistic mechanisms between pollutants and improving the accuracy of composite pollution identification. By tracking pollution migration paths in real time through the dynamic pollution map and combining multi-pollutant relationship data, the invention enables proactive intervention in pollutant interaction responses through treatment solutions, reducing treatment costs while controlling the incidence of secondary pollution.

[0016] Preferably, step S1 includes the following steps:

[0017] S11. Deploy a multi-parameter sensor network in and around the pollution source; the multi-parameter sensor network includes gas sensors, particulate matter monitors, environmental parameter sensors, and operating condition sensors;

[0018] S12. Data is collected through a multi-parameter sensor network to obtain multi-source heterogeneous data; the multi-source heterogeneous data is spatiotemporally aligned and normalized to eliminate noise interference and obtain real-time multi-pollutant data, wherein the real-time multi-pollutant data includes data of multiple real-time pollutants.

[0019] S13. Set a multi-pollutant concentration threshold set; the multi-pollutant concentration threshold set includes thresholds for the concentrations of various pollutants;

[0020] S14. Based on the real-time multi-pollutant data, obtain a real-time multi-pollutant concentration set; the real-time multi-pollutant concentration set contains the real-time concentration of each pollutant.

[0021] S15. Based on the multi-pollutant concentration threshold set, detect whether there are pollutants with a concentration ≥ pollutant concentration threshold in the real-time multi-pollutant concentration set; if not, obtain real-time multi-pollutant data that poses no danger; if so, trigger an alarm and directly treat pollutants with a concentration ≥ pollutant concentration threshold until the pollutant concentrations in the real-time multi-pollutant concentration set are all less than the corresponding pollutant concentration threshold, and obtain real-time multi-pollutant data that poses no danger.

[0022] The above steps achieve comprehensive coverage and dynamic perception of pollution monitoring by constructing a multi-parameter sensor network; by deploying a multi-parameter sensor network in the core area of ​​pollution sources and diffusion paths, a multi-dimensional data acquisition system of "pollution emission-environmental diffusion" is formed; the spatiotemporal alignment algorithm solves the difference in sensor sampling frequency, Z-score normalization eliminates the influence of dimensions, and wavelet denoising technology filters out electromagnetic interference and other noise, ultimately generating a real-time multi-pollutant data stream with unified spatiotemporal benchmarks and high reliability; pollutant concentration thresholds are set according to national environmental quality standards, industry emission limits, and toxicological studies, and threshold sensitivity is dynamically adjusted for special scenarios; pollutant concentration values ​​at each monitoring point are extracted through a real-time data stream parsing engine to construct a structured concentration matrix; a parallel threshold comparison algorithm is adopted, and when the concentration at a certain point exceeds the threshold, an audible and visual alarm is immediately triggered and the treatment equipment is linked. At the same time, the treatment intensity is dynamically adjusted through a PID control algorithm until the sensor feedback concentration drops to a safe value, forming a closed-loop control loop of "monitoring-alarm-treatment-verification" to ensure that the data entering the subsequent analysis stage all meet the safety baseline.

[0023] Preferably, step S2 includes the following steps:

[0024] S21. Construct a pollutant correlation model using a graph neural network; set the initial learning rate for the initial pollutant correlation model;

[0025] Historical multi-pollutant data is obtained by collecting data on the types, concentrations, and environmental conditions of various pollutants. Relationship data between various pollutants in the historical multi-pollutant data is also collected to obtain historical multi-pollutant relationship data. The historical multi-pollutant data and the historical multi-pollutant relationship data together constitute historical data.

[0026] Historical multi-pollutant data is divided into historical multi-pollutant training data and historical multi-pollutant test data, and historical multi-pollutant relationship data is divided into historical multi-pollutant relationship training label data and historical multi-pollutant relationship test label data.

[0027] The training error threshold of the pollutant correlation model is set as follows: α 1. Training error is α 2 and the maximum number of training iterations is β The pollutant correlation model is trained using historical multi-pollutant training data and historical multi-pollutant relationship training label data; the network structure of the pollutant correlation model is adjusted during training; when the training error... α 2≤training error threshold α 1 or reaching the maximum number of training iterations β At that time, an initial pollutant correlation model was obtained;

[0028] S22. The initial pollutant correlation model is tested using historical multi-pollutant test data and historical multi-pollutant relationship test label data. After the test is completed, the optimized pollutant correlation model is obtained.

[0029] S23. Input the non-hazardous real-time multi-pollutant data into the optimized pollutant correlation model to obtain real-time multi-pollutant relationship data;

[0030] The above steps utilize graph neural networks and multi-source data fusion technology to achieve intelligent analysis of pollutant interaction relationships and optimize governance decisions. A pollutant correlation model is constructed using a graph attention network, with an initial learning rate. Historical monitoring data from environmental protection departments, meteorological data, and process parameters from chemical enterprises are integrated to build a pollution event database. Each known interaction relationship between pollutants is labeled through literature mining and laboratory data. The data is divided into training and testing sets, preserving temporal continuity. A dynamic structure optimization mechanism is introduced during training to modify the number of graph convolutional layers, resulting in an optimized pollutant correlation model. After real-time data is input into the model, the correlation strength between pollutants is calculated using a multi-head attention mechanism, generating real-time multi-pollutant relationship data.

[0031] Preferably, step S22 includes the following steps:

[0032] S221. Set the test accuracy threshold as follows: e 1; The initial pollutant correlation model was tested using the historical multi-pollutant test data and historical multi-pollutant relationship test label data, and the test accuracy was obtained as follows: e 2;

[0033] S222, When testing accuracy e 2≥ test accuracy threshold e At step 1, the initial pollutant correlation model is used as the optimized pollutant correlation model; when the test accuracy... e 2 < Test accuracy threshold e At step 1, an optimization algorithm is used to find the learning rate of the initial pollutant correlation model to obtain the optimal solution; the optimal solution is used as the learning rate of the initial pollutant correlation model to obtain the optimized pollutant correlation model.

[0034] The above steps validated the accuracy of the initial pollutant correlation model, ensuring the accuracy of the final model.

[0035] Preferably, step S222 involves using an optimization algorithm to find the learning rate of the initial pollutant correlation model and obtain the optimal solution, including the following steps:

[0036] S2221. Based on the learning rate of the initial pollutant correlation model, randomly set the initial position of the simplex vertex to obtain the simplex vertex set; define the fitness function of the vertex position.

[0037] S2222: Iterate over the simplex vertex set and combine it with the fitness function to obtain the global optimal vertex position;

[0038] S2223, repeat S2222, stop iterating when the maximum number of optimization iterations is reached, and take the global best vertex position as the optimal solution;

[0039] The above steps utilize the Nelder-Mead simplex method to intelligently optimize the model learning rate; based on the initial model learning rate, an initial simplex vertex set is randomly generated in the hyperparameter space (the number of vertices is determined by the learning rate dimension). n The fitness function is defined as a weighted sum of the accuracy on the test set and the accuracy on the training set. The vertex positions are iteratively optimized through geometric transformation operations such as reflection, expansion, and contraction to obtain the optimal learning rate. GPU parallel acceleration and early stopping strategy are adopted simultaneously to improve the correlation prediction response speed and reduce the overfitting rate in the case of sudden pollution.

[0040] Preferably, step S3 includes the following steps:

[0041] S31. Construct a multi-pollutant structure graph based on the real-time multi-pollutant relationship data; the multi-pollutant structure graph includes nodes and edge weights; the nodes represent pollutant types, and the edge weights represent synergistic or antagonistic relationships.

[0042] S32. Based on the real-time multi-pollutant relationship data and the multi-pollutant structure diagram, obtain real-time key pollutant data; based on the real-time key pollutant data, construct a real-time multi-pollutant treatment list, which includes the order of treatment for each pollutant.

[0043] The above steps improve the efficiency of collaborative governance of multiple pollutants by constructing a multi-pollutant structure diagram based on the real-time multi-pollutant relationship data and identifying key pollutants to further generate a treatment list.

[0044] Preferably, step S4 includes the following steps:

[0045] S41. Set constraints, including equipment safety constraints, emission compliance constraints, and resource limitation constraints;

[0046] A multi-objective function is constructed based on the comprehensive pollutant removal rate, cost, and by-product generation. The objectives of the multi-objective function are to maximize the pollutant removal rate, minimize cost consumption, and minimize by-product generation.

[0047] Based on the aforementioned constraints and a real-time multi-pollutant treatment inventory, an initial set of treatment schemes is constructed; the initial set of treatment schemes includes... h A governance scheme, wherein the governance scheme includes multiple control parameters;

[0048] Vectorize the control parameters of the governance schemes in the initial governance scheme set to obtain the initial governance scheme vector set;

[0049] S42. Construct a chromosome population based on the initial governance scheme vector set, and set the maximum number of selection iterations; each chromosome represents a vectorized governance scheme.

[0050] S43. Use the multi-objective function as the fitness function of the chromosome population; perform iterative operations on the chromosome population; calculate the fitness function value of each chromosome according to the fitness function; perform selection, crossover, and mutation operations on the chromosomes in the chromosome population according to the fitness function value of each chromosome to obtain the operated chromosome population;

[0051] S44. Repeat S43. When the maximum number of selection iterations is reached, stop the iteration and obtain the set of optimized solutions.

[0052] S45. Select the optimal solution from the set of optimization solutions using the entropy weight TOPSIS method; construct a dynamic pollution map based on real-time multi-pollutant data without danger, combined with geographic information system and time series analysis; the dynamic pollution map is used to mark pollution hotspots and migration paths;

[0053] The above steps establish constraints, such as equipment safety, emission compliance, and resource limitations, and construct a multi-objective function to maximize pollutant removal rate, minimize cost, and minimize byproduct generation. Combining a real-time multi-pollutant treatment inventory and constraints, an initial set of treatment schemes is constructed, containing multiple treatment options, and the control parameters of each scheme are vectorized. These vectorized treatment schemes are used to construct a chromosome population, with a maximum number of iterations set. The multi-objective function is used as the fitness function for iterative optimization, continuously optimizing the chromosome population through selection, crossover, and mutation operations. When the maximum number of iterations is reached, iteration stops, resulting in an optimized scheme set. The optimal scheme is selected from the optimized scheme set using the entropy-weighted TOPSIS method. A pollutant concentration heatmap is generated based on spatial interpolation technology from a geographic information system. A time-series prediction model is overlaid to analyze pollution diffusion trends, dynamically generating a dual-dimensional pollution map integrating "spatial distribution and temporal evolution." This map can label high-concentration hotspots in real time and simulate migration paths. Control commands are then generated to achieve coordinated treatment of multiple pollutants. This effectively integrates multi-objective optimization and the entropy-weighted TOPSIS method, aiming to achieve efficient, economical, and environmentally friendly pollutant treatment.

[0054] Preferably, step S5 includes the following steps:

[0055] S51. Based on the optimal solution and dynamic pollution map, control commands are generated to coordinate the treatment of multiple pollutants.

[0056] S52. Analyze the composition of byproducts, including waste liquid, waste residue, and waste gas, generated during direct treatment and collaborative treatment processes, and construct a material flow network; the material flow network is marked with recyclable resources; and the recyclable resources are reused according to the material flow network.

[0057] The above steps achieve closed-loop optimization of pollution control by using material flow network analysis and targeted recycling technology in the process of effectively treating multiple pollutants.

[0058] A safety and environmental protection management system based on the collaborative governance of multiple pollutants is used to implement the aforementioned safety and environmental protection management method based on the collaborative governance of multiple pollutants. The system includes a real-time multi-pollutant data acquisition module, a pollutant hazard threshold detection and emergency treatment module, a pollutant correlation modeling and treatment list generation module, a multi-objective optimization and collaborative governance scheme generation module, and a by-product resource utilization and recycling module.

[0059] The real-time multi-pollutant data acquisition and dynamic pollution map construction module collects pollution source and surrounding environmental data in real time by deploying a multi-parameter sensor network, transmits the data using the Internet of Things protocol and performs spatiotemporal alignment and normalization processing to eliminate noise and obtain real-time multi-pollutant data.

[0060] The pollutant hazard threshold detection and emergency treatment module is based on a preset set of multi-pollutant concentration thresholds. It compares and detects real-time monitoring data item by item. When the pollutant concentration exceeds the standard, it triggers a graded alarm mechanism and starts targeted treatment equipment for direct intervention. Through closed-loop control, it ensures that the pollutant concentration drops below the safety threshold and finally outputs standardized multi-pollutant data with no danger.

[0061] The pollutant correlation modeling and treatment list generation module uses a graph neural network to construct a pollutant interaction model. It is trained with historical data and the Nelder-Mead simplex method is used to optimize the model learning rate. The model can analyze the synergistic / antagonistic relationships between pollutants and finally generate a multi-pollutant structure graph containing node weights and correlation strength. Based on this, key pollutants are identified and a real-time multi-pollutant treatment list with treatment priorities is formulated.

[0062] The multi-objective optimization and collaborative governance scheme generation module is used to construct a multi-objective optimization model that includes equipment safety, emission compliance and resource constraints, with the objective function being to maximize the removal rate and minimize the cost and by-products. It uses a genetic algorithm to iteratively optimize the governance scheme vector, combines the entropy weight TOPSIS method to select the optimal scheme from the set of optimized schemes, and combines dynamic pollution maps to generate executable control instructions to achieve collaborative removal of multiple pollutants.

[0063] The by-product resource utilization and recycling module is used to analyze the composition of waste liquid, waste residue and waste gas generated in the treatment process through material flow analysis technology, and establish a resource recycling network; it uses technologies such as electrochemical deposition, waste heat conversion and material modification to realize heavy metal extraction, waste heat power generation and solid waste resource utilization, forming a closed-loop material cycle system of "treatment-recycling-reuse".

[0064] (III) Beneficial Effects

[0065] The present invention has the following beneficial effects:

[0066] This invention integrates multi-source data fusion, deep learning algorithm collaborative optimization, and closed-loop safety management technologies to construct a dynamic pollution map and an optimized pollutant correlation model. This effectively addresses the limitations of traditional pollutant treatment methods, such as singular treatment approaches, removal rate-oriented limitations, and improper byproduct disposal. Based on a multi-parameter sensor network, the system integrates historical data and intelligent optimization algorithms to construct a pollutant correlation model, accurately analyzing the synergistic / antagonistic mechanisms between pollutants and improving the accuracy of composite pollution identification. By tracking pollution migration paths in real time through the dynamic pollution map and combining multi-pollutant relationship data, the invention enables proactive intervention in pollutant interaction responses through treatment solutions, reducing treatment costs while controlling the incidence of secondary pollution.

[0067] This invention employs a multi-objective optimization algorithm to construct a balanced system between environmental and economic benefits. By setting a three-dimensional objective function that includes removal rate, cost, and by-product quantity, and using an improved genetic algorithm for multi-objective optimization, it breaks through the limitations of traditional single removal rate-oriented optimization. The entropy weight TOPSIS optimization mechanism can quantitatively evaluate the comprehensive benefits of the scheme, reduce treatment costs, reduce equipment consumption, and control the amount of by-product generation.

[0068] This invention establishes a comprehensive resource recovery system for by-products, achieving closed-loop optimization of environmental governance. By-product identification technology based on material flow network analysis can accurately identify recyclable components. Through targeted separation and reprocessing technologies, it can improve recycling efficiency while increasing additional revenue, realizing the transformation of pollution control from "end-of-life treatment" to a "circular economy."

[0069] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0070] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, the drawings can be obtained from these drawings without creative effort.

[0071] Figure 1 This is a schematic diagram of a safety and environmental management method based on the synergistic treatment of multiple pollutants according to the present invention.

[0072] Figure 2 This is a flowchart illustrating the process of obtaining an optimized pollutant correlation model in a safety and environmental management method based on multi-pollutant synergistic governance according to the present invention.

[0073] Figure 3 This is a schematic diagram of the process for obtaining a real-time multi-pollutant treatment list in a safety and environmental management method based on multi-pollutant synergistic treatment according to the present invention.

[0074] Figure 4 This is a flowchart illustrating the process of obtaining the optimal solution in a safety and environmental management method based on the synergistic treatment of multiple pollutants, as described in this invention.

[0075] Figure 5 This is a schematic diagram of a safety and environmental protection management system based on the collaborative treatment of multiple pollutants according to the present invention. Detailed Implementation

[0076] The technical solutions of the embodiments of the invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the invention, and not all embodiments. Based on the embodiments of the invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the invention.

[0077] In the description of this invention, it should be understood that the terms "opening", "upper", "lower", "top", "middle", "inner", etc., which indicate orientation or positional relationship, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the invention.

[0078] Example 1

[0079] Please see Figure 1 , Figure 2 , Figure 3 , Figure 4This invention discloses a safety and environmental management method based on the synergistic treatment of multiple pollutants, comprising the following steps:

[0080] S1. Detect multiple pollutants to obtain real-time multi-pollutant data;

[0081] The pollutants corresponding to the real-time pollutant data that are dangerous are directly processed so that all real-time pollutants are no longer dangerous, resulting in dangerous real-time multi-pollutant data.

[0082] S1 includes the following steps:

[0083] S11. Deploy a multi-parameter sensor network at pollution sources (such as industrial chimneys and sewage discharge outlets) and their surrounding environment; the multi-parameter sensor network includes gas sensors, particulate matter monitors, environmental parameter sensors, and operating condition sensors; the gas sensors are used to monitor the concentrations of various gases such as SO2, NOx, VOCs, and O3 in real time; the particulate matter monitors are used to detect the content of various particulate matter such as PM2.5, PM10, and heavy metals (such as lead and cadmium); the environmental parameter sensors are used to collect various environmental data such as temperature, humidity, air pressure, and wind speed; the operating condition sensors are used to monitor equipment operating parameters (such as reactor pressure, flow rate, and temperature); the sensor data is transmitted via Internet of Things protocols (such as MQTT and LoRa);

[0084] S12. Data is collected through a multi-parameter sensor network to obtain multi-source heterogeneous data; the multi-source heterogeneous data is spatiotemporally aligned and normalized to eliminate noise interference and obtain real-time multi-pollutant data, wherein the real-time multi-pollutant data includes data of multiple real-time pollutants.

[0085] S13. Set a set of multi-pollutant concentration thresholds. ,in, j i Indicates the concentration threshold of multiple pollutants. i Concentration thresholds for various pollutants k This represents the total number of pollutant types; the multi-pollutant concentration threshold set includes thresholds for the concentrations of various pollutants.

[0086] S14. Based on the real-time multi-pollutant data, obtain the real-time multi-pollutant concentration set. ,in, l i Indicates the real-time concentration of multiple pollutants. i The concentration of the pollutants, p This represents the total number of pollutant types in real time; the real-time multi-pollutant concentration set contains the real-time concentration of each pollutant.

[0087] S15. Based on the multi-pollutant concentration threshold set, detect whether there are pollutants with a concentration ≥ pollutant concentration threshold in the real-time multi-pollutant concentration set; if not, obtain real-time multi-pollutant data that poses no danger; if so, trigger an alarm and directly treat pollutants with a concentration ≥ pollutant concentration threshold until the pollutant concentrations in the real-time multi-pollutant concentration set are all less than the corresponding pollutant concentration threshold, and obtain real-time multi-pollutant data that poses no danger.

[0088] S2. Input the real-time multi-pollutant data that poses no danger into the optimized pollutant correlation model obtained by combining historical data with optimization algorithms to obtain real-time multi-pollutant relationship data;

[0089] S2 includes the following steps:

[0090] S21. Construct a pollutant correlation model using a graph neural network; set the initial learning rate for the initial pollutant correlation model;

[0091] Historical multi-pollutant data is obtained by collecting data on the types, concentrations, and environmental conditions of various pollutants. Relationship data between various pollutants in the historical multi-pollutant data is also collected to obtain historical multi-pollutant relationship data. The historical multi-pollutant data and the historical multi-pollutant relationship data together constitute historical data.

[0092] Historical multi-pollutant data is divided into historical multi-pollutant training data and historical multi-pollutant test data, and historical multi-pollutant relationship data is divided into historical multi-pollutant relationship training label data and historical multi-pollutant relationship test label data.

[0093] The training error threshold of the pollutant correlation model is set as follows: α 1. Training error is α 2 and the maximum number of training iterations is β The pollutant correlation model is trained using historical multi-pollutant training data and historical multi-pollutant relationship training label data; the network structure of the pollutant correlation model is adjusted during training; when the training error... α 2≤training error threshold α 1 or reaching the maximum number of training iterations β At that time, an initial pollutant correlation model was obtained;

[0094] S22. The initial pollutant correlation model is tested using historical multi-pollutant test data and historical multi-pollutant relationship test label data. After the test is completed, the optimized pollutant correlation model is obtained.

[0095] S22 includes the following steps:

[0096] S221. Set the test accuracy threshold as follows: e 1; The initial pollutant correlation model was tested using the historical multi-pollutant test data and historical multi-pollutant relationship test label data, and the test accuracy was obtained as follows: e 2;

[0097] S222, When testing accuracy e 2≥ test accuracy threshold e At step 1, the initial pollutant correlation model is used as the optimized pollutant correlation model; when the test accuracy... e 2 < Test accuracy threshold e At step 1, an optimization algorithm is used to find the learning rate of the initial pollutant correlation model to obtain the optimal solution; the optimal solution is used as the learning rate of the initial pollutant correlation model to obtain the optimized pollutant correlation model.

[0098] The steps in S222 to find the learning rate of the initial pollutant correlation model using an optimization algorithm to obtain the optimal solution include the following:

[0099] S2221. Based on the learning rate dimension of the initial pollutant correlation model. n , generated by n A simplex consisting of +1 vertices (the number of vertices is fixed). n +1 (no need to set population size parameter); set the maximum number of optimization iterations as the termination condition;

[0100] The initial set of vertex positions of the simplex is randomly set as follows: The position of each vertex e i This represents a candidate solution with a learning rate.

[0101] According to the test accuracy threshold e 1 and training accuracy e 2. Define the fitness function for the vertex position. The fitness function formula is as follows:

[0102] ;

[0103] d Represents the fitness function. c Indicates the bias amount;

[0104] S2222. Calculate the fitness value of each vertex and sort them from high to low fitness to obtain the initial fitness set. e best ≥ e good ≥...≥ e worst Find the worst-case scenario. e worstFind the centroids of all external vertices to obtain the mean centroid of the vertices;

[0105] Based on the mean of vertex centroid and e worst The reflection point is obtained. e r ;like d ( e r > d ( e good ), then use e r replace e worst ;like d ( e r > d ( e best ), then based on the reflection point e r Generate extension points by taking the average of the centroids of the vertices. e p ;like d ( e p > d ( e r ), then use e p replace e worst ;

[0106] like d ( e r )≤ d ( e good If the reflection fails, it is determined to be a failure, based on... e worst The contraction point is generated by the mean of the centroid of the vertex. e q ;like d ( e q > d ( e worst ), then use e q replace e worst ;

[0107] like d ( e q )≤ d ( e worstIf the contraction fails, the process moves to the optimal vertex. e best Compress all vertices and record the globally best vertex; update the historical best vertex in each iteration.

[0108] S2223, repeat S2222, stop iterating when the maximum number of optimization iterations is reached, and take the global best vertex as the optimal solution;

[0109] S23. Input the non-hazardous real-time multi-pollutant data into the optimized pollutant correlation model to obtain real-time multi-pollutant relationship data;

[0110] S3. Construct a real-time multi-pollutant treatment list based on real-time multi-pollutant relationship data;

[0111] S3 includes the following steps:

[0112] S31. Construct a multi-pollutant structure graph based on the real-time multi-pollutant relationship data; the multi-pollutant structure graph includes nodes and edge weights; the nodes represent pollutant types, and the edge weights represent synergistic or antagonistic relationships (such as VOCs and NOx generating ozone under light).

[0113] S32. Based on the real-time multi-pollutant relationship data and multi-pollutant structure diagram, obtain real-time key pollutant data (e.g., PM2.5 in a certain area mainly originates from the secondary reaction of SO2 and NH3); based on the real-time key pollutant data, construct a real-time multi-pollutant treatment list, which includes the order of treatment for each pollutant;

[0114] S4. Based on the constraints and the real-time multi-pollutant treatment list, an initial set of treatment schemes is obtained;

[0115] The optimal governance scheme is obtained by combining the initial governance scheme set with an optimization algorithm;

[0116] S4 includes the following steps:

[0117] S41. Set constraints, including equipment safety constraints, emission compliance constraints, and resource limitation constraints;

[0118] A multi-objective function is constructed based on the overall pollutant removal rate, cost, and byproduct generation. The objectives of this multi-objective function are to maximize the pollutant removal rate, minimize cost, and minimize byproduct generation. The formula for the multi-objective function is as follows.

[0119] ;

[0120] in, F 1. F 2 and F3 represents the weighted sum of the overall pollutant removal rate, the weighted sum of costs, and the weighted sum of by-product generation, respectively. n This indicates the total number of pollutant types. w i Indicates the set number i The weighting coefficients for each pollutant (dynamically adjusted based on factors such as the pollutant's toxicity and emission limits), or i Indicates the first i Removal rate of various pollutants d 1 represents the weighting coefficient for the total energy consumption of the equipment. E This indicates the total energy consumption of the equipment. d 2 indicates the weighting coefficient for setting the amount of medicine consumed. C Indicates the amount of medicine consumed, o represents the total number of types of byproducts, and vi represents the set number of byproducts. i Impact factor of a byproduct (used to indicate the degree of environmental impact of the byproduct). Q i Indicates the first i The amount of by-products generated;

[0121] Based on the aforementioned constraints and the real-time multi-pollutant treatment inventory, an initial set of treatment schemes is constructed. ,in f i Represents the first in the initial governance scheme set. i One governance plan, h This indicates the total number of initial governance schemes, each of which includes multiple control parameters;

[0122] The initial governance scheme set The control parameters are vectorized to obtain the initial governance scheme vector set. ,in g i Represents the vectorized first... i One governance solution, such as g i =[Amount of activated carbon, reaction temperature, concentration of alkali solution];

[0123] S42. Construct a chromosome population based on the initial governance scheme vector set, then the chromosome population is: , m i Represents the first chromosome in a population. i There are 3 chromosomes, with a maximum number of selection iterations set; each chromosome represents a vectorized governance scheme.

[0124] S43. Use the multi-objective function as the fitness function of the chromosome population; perform iterative operations on the chromosome population; calculate the fitness function value of each chromosome according to the fitness function; perform selection, crossover, and mutation operations on the chromosomes in the chromosome population according to the fitness function value of each chromosome to obtain the operated chromosome population;

[0125] S44. Repeat S43. When the maximum number of selection iterations is reached, stop the iteration and obtain the set of optimized solutions.

[0126] S45. Select the optimal solution from the set of optimization solutions using the entropy weight TOPSIS method; construct a dynamic pollution map based on real-time multi-pollutant data without danger, combined with geographic information system and time series analysis; the dynamic pollution map is used to mark pollution hotspots and migration paths;

[0127] S5. Based on the optimal solution and dynamic pollution map, conduct synergistic treatment of multiple pollutants;

[0128] S5 includes the following steps:

[0129] S51. Based on the optimal solution and dynamic pollution map, control commands are generated to coordinate the treatment of multiple pollutants.

[0130] S52. Analyze the composition of byproducts, including waste liquid, waste residue, and waste gas, generated during direct treatment and collaborative treatment processes, and construct a material flow network; the material flow network is labeled with recyclable resources (such as heavy metals in waste liquid and waste heat from the reaction).

[0131] Based on the material flow network, recyclable resources are reused; for example, for heavy metals, copper, zinc and other metals are extracted from waste liquid by electrochemical deposition, the waste heat from the reaction is converted into steam for heating or power generation in the plant area through a heat exchanger, and desulfurized gypsum is processed into building materials or used as a soil conditioner.

[0132] Example 2

[0133] Please see Figure 5 A safety and environmental protection management system based on the collaborative governance of multiple pollutants is used to implement the above-mentioned safety and environmental protection management method based on the collaborative governance of multiple pollutants. It includes a real-time multi-pollutant data acquisition module, a pollutant hazard threshold detection and emergency treatment module, a pollutant correlation modeling and treatment list generation module, a multi-objective optimization and collaborative governance scheme generation module, and a by-product resource utilization and recycling module.

[0134] The real-time multi-pollutant data acquisition and dynamic pollution map construction module collects pollution source and surrounding environmental data in real time by deploying a multi-parameter sensor network, transmits the data using the Internet of Things protocol and performs spatiotemporal alignment and normalization processing to eliminate noise and obtain real-time multi-pollutant data.

[0135] The pollutant hazard threshold detection and emergency treatment module is based on a preset set of multi-pollutant concentration thresholds. It compares and detects real-time monitoring data item by item. When the pollutant concentration exceeds the standard, it triggers a graded alarm mechanism and starts targeted treatment equipment for direct intervention. Through closed-loop control, it ensures that the pollutant concentration drops below the safety threshold and finally outputs standardized multi-pollutant data with no danger.

[0136] The pollutant correlation modeling and treatment list generation module uses a graph neural network to construct a pollutant interaction model. It is trained with historical data and the Nelder-Mead simplex method is used to optimize the model learning rate. The model can analyze the synergistic / antagonistic relationships between pollutants and finally generate a multi-pollutant structure graph containing node weights and correlation strength. Based on this, key pollutants are identified and a real-time multi-pollutant treatment list with treatment priorities is formulated.

[0137] The multi-objective optimization and collaborative governance scheme generation module is used to construct a multi-objective optimization model that includes equipment safety, emission compliance and resource constraints, with the objective function being to maximize the removal rate and minimize the cost and by-products; it uses a genetic algorithm to iteratively optimize the governance scheme vector, and combines the entropy weight TOPSIS method to select the optimal scheme from the set of optimized schemes, generating executable control instructions to achieve the collaborative removal of multiple pollutants;

[0138] The by-product resource utilization and recycling module is used to analyze the composition of waste liquid, waste residue and waste gas generated in the treatment process through material flow analysis technology, and establish a resource recycling network; it uses technologies such as electrochemical deposition, waste heat conversion and material modification to realize heavy metal extraction, waste heat power generation and solid waste resource utilization, forming a closed-loop material cycle system of "treatment-recycling-reuse".

[0139] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0140] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A safety and environmental management method based on the synergistic treatment of multiple pollutants, characterized in that, Includes the following steps: S1. Detect multiple pollutants to obtain real-time multi-pollutant data; The pollutants corresponding to the real-time pollutant data that are dangerous are directly processed to make all real-time multi-pollutants non-dangerous, resulting in non-dangerous real-time multi-pollutant data. S2. Input the real-time multi-pollutant data that poses no danger into the optimized pollutant correlation model obtained by combining historical data with optimization algorithms to obtain real-time multi-pollutant relationship data; S2 includes the following steps: S21. The pollutant correlation model is trained using the historical multi-pollutant training data and historical multi-pollutant relationship training label data in the historical data to obtain the initial pollutant correlation model. S22. The initial pollutant correlation model is tested using historical multi-pollutant test data and historical multi-pollutant relationship test label data in the historical data to obtain an optimized pollutant correlation model. S23. Input the non-hazardous real-time multi-pollutant data into the optimized pollutant correlation model to obtain real-time multi-pollutant relationship data; S3. Construct a real-time multi-pollutant treatment list based on real-time multi-pollutant relationship data; S4. Based on the constraints and the real-time multi-pollutant treatment list, an initial set of treatment schemes is obtained; The optimal governance scheme is obtained by combining the initial governance scheme set with an optimization algorithm; S5. Based on the optimal solution and dynamic pollution map, conduct synergistic treatment of multiple pollutants; S5 includes the following steps: S51. Based on the optimal solution and dynamic pollution map, control commands are generated to coordinate the treatment of multiple pollutants; the dynamic pollution map is constructed based on geographic information system and time series analysis combined with real-time multi-pollutant data without danger; the dynamic pollution map is used to mark pollution hotspots and migration paths. S52. Analyze the composition of byproducts generated during direct and collaborative governance processes to construct a material flow network; the material flow network is marked with recyclable resources; and the recyclable resources are reused based on the material flow network.

2. The safety and environmental management method based on the synergistic treatment of multiple pollutants according to claim 1, characterized in that, The presence of danger posed by the real-time multi-pollutant data is determined based on the pollution concentration.

3. The safety and environmental management method based on the synergistic treatment of multiple pollutants according to claim 1, characterized in that, S22 includes the following steps: S221. Set the test accuracy threshold; use the historical multi-pollutant test data and historical multi-pollutant relationship test label data to test the initial pollutant correlation model and obtain the test accuracy. S222. When the test accuracy is less than the test accuracy threshold, find the learning rate of the initial pollutant correlation model to obtain the optimal solution; use the optimal solution as the learning rate of the initial pollutant correlation model to obtain the optimized pollutant correlation model.

4. The safety and environmental management method based on the synergistic treatment of multiple pollutants according to claim 3, characterized in that, S222 includes the following steps: S2221. Based on the learning rate of the initial pollutant correlation model, randomly set the initial positions of the simplex vertices to obtain the simplex vertex set; define the fitness function for the vertex positions. S2222: Iterate over the simplex vertex set and combine it with the fitness function to obtain the global optimal vertex position; S2223, repeat S2222. When the maximum number of optimization iterations is reached, stop the iteration and take the global best vertex position as the optimal solution.

5. A safety and environmental management method based on the synergistic treatment of multiple pollutants according to claim 1, characterized in that, S3 includes the following steps: S31. Construct a multi-pollutant structure graph based on the real-time multi-pollutant relationship data; the multi-pollutant structure graph includes nodes and edge weights; the nodes represent pollutant types, and the edge weights represent synergistic or antagonistic relationships. S32. Based on the real-time multi-pollutant relationship data and multi-pollutant structure diagram, after obtaining the real-time key pollutant data, construct a real-time multi-pollutant treatment list.

6. The safety and environmental management method based on the synergistic treatment of multiple pollutants according to claim 1, characterized in that, S4 includes the following steps: S41. Set constraints and, in conjunction with a real-time multi-pollutant treatment inventory, construct an initial set of treatment schemes; construct a multi-objective function; Vectorize the control parameters of the governance schemes in the initial governance scheme set to obtain the initial governance scheme vector set; S42. Construct a chromosome population based on the initial governance scheme vector set, and set the maximum number of selection iterations; S43. Perform iterative operations on the chromosome population, and perform selection, crossover, and mutation operations on the chromosomes according to the multi-objective function to obtain the chromosome population under operation; S44. Repeat S43. When the maximum number of selection iterations is reached, stop the iteration and obtain the set of optimized solutions. S45. Select the optimal solution from the set of optimized solutions.

7. A safety and environmental protection system based on the synergistic treatment of multiple pollutants, characterized in that, To achieve a safety and environmental management method based on the synergistic treatment of multiple pollutants as described in any one of claims 1-6.

8. A storage medium, characterized in that, It stores a program that, when executed by a processor, implements a safety and environmental management method based on the synergistic treatment of multiple pollutants as described in any one of claims 1-6.

Images