Rapid traceability system for atmospheric pollutants based on big data

Abstract

The invention relates to the technical field of environmental monitoring, and discloses an atmospheric pollutant rapid traceability system based on big data, and the system comprises a diffusion simulation module, a hybrid optimization module, a dynamic algorithm selection module, a multi-source analysis module, and a verification and output module. By constructing a dynamic matching rule of meteorological data characteristics and a traceability algorithm, an anti-interference algorithm model is adaptively switched according to different meteorological conditions, meteorological refraction deviation in pollutant concentration field simulation is eliminated in real time, intensity inversion errors caused by optical interference in a traditional method are reduced, physical authenticity of source intensity estimation is guaranteed, and the method is suitable for large-scale popularization and application. A hybrid optimization module is arranged, the search path state of a genetic algorithm is monitored in real time by using an axial detection and mode movement strategy, parameter space deviation of a mathematical level is automatically identified, a mode search deviation correction mechanism and a genetic algorithm restart operation are triggered, and geometric accuracy and spatial reliability of a traceability result are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of environmental monitoring, in particular to a big data-based rapid atmospheric pollutant tracing system. BACKGROUND

[0002] Atmospheric pollutants refer to those substances discharged into the atmosphere due to human activities and natural processes and having harmful effects on people and the environment. In clean air, the composition of constant gases is negligible, but in a certain range of air, the presence of trace substances in terms of quantity and duration can have adverse effects and hazards on people, animals, plants, and objects and materials. When the concentration of atmospheric pollutants reaches a harmful level, it can destroy the ecological system and the conditions for normal survival and development of human beings, causing harm to people and objects, which is called air pollution.

[0003] Currently, in the field of big data-based atmospheric pollution tracing, traditional tracing systems rely on fixed parameterized models to reverse the location and emission intensity of pollution sources. However, there are significant defects in actual application: due to the dynamic changes of the meteorological field and the spatial heterogeneity of the underlying surface characteristics, the diffusion model cannot real-time perceive and calibrate the meteorological refraction deviation on the light transmission path, resulting in systematic color difference in the simulation of pollutant concentration field. This optical interference poses a risk of intensity misjudgment in the tracing result. Meanwhile, during the execution of the optimization algorithm, there is no dynamic feedback mechanism for the search direction of the pollution source. When the genetic algorithm falls into local optimization due to initial population deviation and fitness deception, it cannot real-time identify and correct the path deviation of the mathematical search angle, causing coordinate positioning error exceeding 500 meters without automatic correction ability. In addition, in the face of the multi-point source superposition scenario of industrial parks, the existing system lacks the hierarchical analysis capability of multiple pollution areas, and cannot achieve confidence interval hierarchical cutting of each independent source intensity in the superimposed concentration field, resulting in confusion between high-contribution source and low-contribution source emission intensity, reducing the precision and effectiveness of responsibility tracing.

[0004] Therefore, the present application proposes a big data-based rapid atmospheric pollutant tracing system to solve the above problems. SUMMARY

[0005] (I) Technical problems solved

[0006] In view of the deficiencies of the prior art, the present application provides a big data-based rapid atmospheric pollutant tracing system to solve the problems raised in the background art.

[0007] (II) Technical solutions

[0008] To achieve the above purposes, the present application provides the following technical solutions: a big data-based rapid atmospheric pollutant tracing system, comprising:

[0009] a diffusion simulation module configured to build a Gaussian diffusion model algorithm and calculate a pollutant concentration field distribution according to input meteorological parameters and underlying surface characteristic parameters;

[0010] a hybrid optimization module coupled to the diffusion simulation module and configured to iteratively optimize pollutant source parameters using a genetic-pattern search hybrid algorithm, including:

[0011] a parameter encoding unit configured to convert the pollutant source parameters into genotypic individuals through decimal encoding;

[0012] a population evolution unit configured to update the source parameter population through selection, crossover and mutation operations, and filter high fitness individuals based on a fitness function;

[0013] a pattern search unit configured to perform axial and pattern movement searches on local optimal solutions output by the genetic algorithm to achieve global optimization;

[0014] a dynamic algorithm selection module configured to match an optimal source tracing algorithm type from a preset algorithm library according to meteorological data characteristics and an observation point arrangement scheme;

[0015] a multi-source analysis module configured to process a multi-point source scenario in parallel and output emission intensity confidence intervals of each pollutant source;

[0016] a verification and output module configured to verify model accuracy based on simulation data and Prairie Grass field test data, and output source tracing results and convergence indicators.

[0017] Preferably, the diffusion simulation module further includes:

[0018] a reliability verification unit configured to verify the reliability of the Gaussian diffusion model based on standardized mean error, standardized mean deviation and root mean square error indicators;

[0019] a parameter calibration unit configured to calibrate model parameters through measured concentration data of the Prairie Grass project to ensure the simulation accuracy of high-value and distribution areas of pollutants.

[0020] Preferably, the fitness function of the hybrid optimization module is defined as:

[0021]

[0022] wherein C 模拟,i is a simulated concentration value of the nth observation point, C 观测,i is a measured concentration value of the nth observation point, and n is the total number of observation points.

[0023] Preferably, the operation of the parameter encoding unit includes:

[0024] a key emission source information database is built, and an initial population is generated by randomly extracting individuals;

[0025] The error between the simulated value and the observed value is compared by parallel computing, and the individuals with fitness higher than the threshold value are screened into the next generation evolution.

[0026] Preferably, the dynamic algorithm selection module matches the algorithm according to the following rules:

[0027] When the observation point is an arc-shaped scheme, the near-source sparse point strategy is preferentially selected;

[0028] When the observation point is a rectangular grid scheme, the near-source dense point strategy is preferentially selected;

[0029] When the atmospheric stability is unstable, the weight of the observation point beyond 400 meters is increased;

[0030] When the atmospheric stability is stable, the weight of the observation point within 200 meters is increased.

[0031] Preferably, the verification and output module includes the following verification processes:

[0032] Simulation evaluation: based on the Prairie Grass meteorological conditions, an ideal concentration field is generated to test the success rate and timeliness of the source tracing under different point distribution schemes, point distribution densities and stability conditions;

[0033] Field verification: the coordinate error distance and intensity deviation of the source tracing result and the real source parameters are compared to quantify the model uncertainty.

[0034] Preferably, the following steps are included:

[0035] S1. Input meteorological data and observation point pollutant concentration data;

[0036] S2. Calculate the concentration field distribution of each source parameter individual by the diffusion simulation module;

[0037] S3. Calculate the fitness value of the population individual based on the error between the observed concentration and the simulated concentration;

[0038] S4. Perform genetic algorithm operations:

[0039] Selection operation: retain the top 30% of individuals in terms of fitness;

[0040] Cross operation: exchange genes between the remaining individuals, and the exchange ratio is set to 60%;

[0041] Mutation operation: randomly mutate individual gene values with a probability of 5%;

[0042] S5. When the genetic algorithm iteration reaches 1000 times, input the current optimal individual into the pattern search algorithm:

[0043] Set the initial step size δ0=0.1 and the contraction factor α=0.5;

[0044] The search range is dynamically reduced by the step update formula:

[0045] delta k+1 = alpha * delta k

[0046] wherein delta k is the search step of the kth iteration, alpha is the step attenuation coefficient, and k is the iteration number index;

[0047] S6. When the error is less than 0.01 and the iteration reaches 2000 times, output the global optimal source parameter (Q, x, y).

[0048] Preferably, before the S1 step, it further comprises:

[0049] The observation point strategy is dynamically adjusted:

[0050] If the atmospheric stability is C level, 50% of the observation points are arranged in the area 400 meters away from the source;

[0051] If the atmospheric stability is E level, 70% of the observation points are arranged in the area within 200 meters from the source.

[0052] Preferably, the specific implementation of the cross operation in S4 is:

[0053] The single-point crossover method is adopted, and a cutting point in the gene sequence is randomly selected to exchange the gene fragments after the cutting point of the parent individual;

[0054] After generating the offspring individual, the fitness is recalculated and the individual below the average is eliminated.

[0055] Preferably, the termination condition of the pattern search in S5 is:

[0056] The step delta k is less than 0.001 and no better solution is found for 50 consecutive iterations;

[0057] Before outputting the final solution, the correlation coefficient between the final solution and the observed concentration is calculated, and if the correlation coefficient is less than 0.95, the genetic algorithm restart mechanism is triggered.

[0058] (Three) beneficial effects

[0059] Compared with the prior art, the present application provides a big data-based rapid atmospheric pollutant tracing system, which has the following beneficial effects:

[0060] 1. In the present application, by setting the dynamic algorithm selection module, when performing atmospheric pollution source tracing calculation, by constructing the dynamic matching rule of meteorological data characteristics and tracing algorithm, the anti-interference algorithm model is switched adaptively for different meteorological conditions, the meteorological refraction deviation in the simulation of the pollutant concentration field is eliminated in real time, the intensity inversion error caused by optical interference in the traditional method is reduced, and the physical authenticity of the source intensity estimation is ensured.

[0061] 2. In the present application, by setting the hybrid optimization module, when performing pollution source parameter search, the search path state of the genetic algorithm is monitored in real time by using the axial detection and mode moving strategy, the parameter space deviation in the mathematical level is automatically identified, and the mode search correction mechanism and the genetic algorithm restart operation are triggered, so that the coordinate positioning deviation risk is solved, and the geometric accuracy and spatial reliability of the tracing result are improved.

[0062] 3. In the present application, by setting the multi-source analysis module, in the multi-point source scene of the industrial park, the independent contribution source of the superimposed concentration field is separated by using the confidence interval hierarchical cutting technology, the emission intensity of the main pollution source and the secondary pollution source is spatially decoupled and interval quantized based on the Bayesian probability model, the problem of confusion of high and low contribution sources in the traditional method is solved, and the accuracy and law enforcement effectiveness of the responsibility tracing are improved. BRIEF DESCRIPTION OF DRAWINGS

[0063] Figure 1 It is the schematic diagram of the overall system framework of the present application. DETAILED DESCRIPTION

[0064] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0065] Please refer to Figure 1 The atmospheric pollutant rapid tracing system based on big data comprises:

[0066] A diffusion simulation module is configured to construct a Gaussian diffusion model algorithm, and calculate the pollutant concentration field distribution according to the input meteorological parameters and underlying surface characteristic parameters.

[0067] A hybrid optimization module is coupled to the diffusion simulation module, and a genetic-mode search hybrid algorithm is used to iteratively optimize the pollution source parameters, comprising:

[0068] A parameter encoding unit is configured to convert the pollution source parameters into genotypes by decimal encoding;

[0069] The population evolution unit updates the source parameter population through selection, crossover and mutation operations, and screens high fitness individuals based on a fitness function;

[0070] The mode searching unit searches the local optimal solution output by the genetic algorithm in the axial and mode moving directions to realize global optimization;

[0071] The dynamic algorithm selection module matches the optimal source tracing algorithm type from the preset algorithm library according to the meteorological data characteristics and the observation point arrangement scheme;

[0072] The multi-source analysis module processes the multi-point source scene in parallel and outputs the emission intensity confidence interval of each pollution source;

[0073] The verification and output module verifies the model accuracy based on simulation data and Prairie Grass field test data, and outputs the source tracing results and convergence indicators.

[0074] The diffusion simulation module further comprises:

[0075] The reliability of the Gaussian diffusion model is verified based on the standardized mean error, standardized mean deviation and root mean square error indicators;

[0076] The model parameters are calibrated through the measured concentration data of the Prairie Grass project to ensure the simulation accuracy of the high-value area and distribution area of pollutants.

[0077] The fitness function of the hybrid optimization module is defined as:

[0078]

[0079] Where C 模拟,i is the simulated concentration value of the i th observation point, C 观测,i is the measured concentration value of the i th observation point, and n is the total number of observation points.

[0080] The operations of the parameter encoding unit include:

[0081] An initial population is generated by randomly selecting individuals from the key emission source information library;

[0082] The individuals with a fitness higher than the threshold value are selected into the next generation evolution by comparing the simulation value and the observation value error through parallel calculation.

[0083] The dynamic algorithm selection module matches the algorithm according to the following rules:

[0084] When the observation point arrangement is an arc scheme, the near-source sparse point arrangement strategy is preferred;

[0085] When the observation point arrangement is a rectangular grid scheme, the near-source dense point arrangement strategy is preferred;

[0086] When the atmospheric stability is unstable, increase the weight of observation points 400 meters away;

[0087] When the atmospheric stability is stable, increase the weight of observation points within 200 meters.

[0088] The verification and output module's verification process includes:

[0089] Simulation evaluation: generate ideal concentration field based on Prairie Grass meteorological conditions, test the success rate and timeliness of different point layout schemes, point layout density, and stability conditions;

[0090] Field verification: compare the coordinate error distance and intensity deviation of the source parameters of the traceability results and the real source parameters to quantify the model uncertainty.

[0091] Including the following steps:

[0092] S1. Input meteorological data and observation point pollutant concentration data;

[0093] S2. Calculate the concentration field distribution of each source parameter individual through the diffusion simulation module;

[0094] S3. Calculate the fitness value of the population individual based on the error between the observed concentration and the simulated concentration;

[0095] S4. Perform genetic algorithm operations:

[0096] Selection operation: retain the top 30% of individuals in terms of fitness;

[0097] Cross operation: exchange genes between the remaining individuals, with a swap ratio of 60%;

[0098] Mutation operation: randomly mutate individual gene values with a probability of 5%;

[0099] S5. When the genetic algorithm iteration reaches 1000 times, input the current optimal individual into the pattern search algorithm:

[0100] Set the initial step size δ0=0.1 and the contraction factor α=0.5;

[0101] Update the search range dynamically through the step size update formula:

[0102] δ k+1 = α·δ k

[0103] Where δ k is the search step size of the kth iteration, α is the step size attenuation coefficient, and k is the iteration number index;

[0104] S6. When the error is less than 0.01 and the iteration reaches 2000 times, output the global optimal source parameters (Q, x, y).

[0105] Before S1 step, also includes:

[0106] Dynamic adjustment of observation point strategy:

[0107] If the atmospheric stability is C level, 50% of the observation points are arranged in the area 400 meters away from the source;

[0108] If the atmospheric stability is E level, 70% of the observation points are arranged in the area within 200 meters from the source.

[0109] The specific implementation of the cross operation in S4 is:

[0110] Single-point crossover method is adopted, and a cutting point in the gene sequence is randomly selected to exchange the gene fragments after cutting the parent individual;

[0111] After generating the offspring individual, the fitness is recalculated and the individual below the average is eliminated.

[0112] The termination condition of the pattern search in S5 is:

[0113] Step size δ k Less than 0.001 and no better solution is found for 50 consecutive iterations;

[0114] Before outputting the final solution, the correlation coefficient between the solution and the observed concentration is calculated, and if it is lower than 0.95, the genetic algorithm restart mechanism is triggered.

[0115] Example one, industrial area pollution event traceability

[0116] A PM2.5 concentration abnormal surge event occurred in a petrochemical park in the early morning. The system monitored that the wind direction fluctuation frequency reached 2 times per minute and the atmospheric stability jumped to E level in real time through the dynamic algorithm selection module. According to the preset rules, the near-field dense point strategy was automatically enabled: 17 mobile monitoring points were deployed within a radius of 200 meters from the center of the pollution source, and the infrared absorption spectrum correction model in the water vapor refractive algorithm library was activated.

[0117] The hybrid optimization module generates an initial population with an initial emission intensity of 500 kg / h and coordinates (X0, Y0). After 1200 genetic iterations, the suspected source is located at the A3 position of the storage tank area. At this time, the pattern search unit detects that the axial deviation is out of limit, and immediately triggers the axial compensation mechanism for 5 times of pattern movement, and the source coordinates are corrected to the valve leakage point of the storage tank (X0+3.2m, Y0-1.7m).

[0118] The multi-source analysis module simultaneously separates the benzene series and hydrogen sulfide superimposed concentration field, and outputs that the main pollution source is the valve leakage and the secondary pollution source is the abnormal emission of chimney No. 3. The final traceability result is compared with the valve repair record and chimney working condition log by the verification module, and the error is less than 1.7%.

[0119] Example Two, Multi-source Industrial Park Responsibility Definition

[0120] In response to the problem of ozone concentration exceeding the standard in a certain chemical cluster area all year round, the system starts a multi-source collaborative tracing mode. The dynamic algorithm selection module identifies that the observation network is arranged in a 2km x 2km rectangular grid layout, and the atmospheric stability is C level unstable state, automatically matches the sparse weighted strategy of far source: 40% of the monitoring points are arranged 800 meters away from the core area, and the distance weight coefficient is assigned as 0.6.

[0121] The diffusion simulation module loads the wind profile parameters obtained from the terrain wind tunnel experiment, and constructs a composite diffusion model containing chloric acid and VOCs. The mixed optimization module first genetically evolves to four potential sources: A factory reaction kettle, B factory tank group, C factory incinerator, D factory loading and unloading area, and then the mode search unit starts independent axial detection for each source: when the fitness function of B factory tank group is detected to have shock, the genetic algorithm restart mechanism is activated to reinitialize the population.

[0122] After 3 rounds of collaborative optimization, the multi-source analysis module uses Bayesian hierarchical sampling to cut out the main and secondary contribution sources: B factory tank leakage contribution 78.5%, intensity interval [2.3t / a, 2.5t / a], C factory incinerator contribution 12.1%, interval [0.35t / a, 0.38t / a], the total contribution of the remaining sources is less than 9%, and the environmental law enforcement department implements precise production limit on B factory according to the results.

[0123] Example Three, Cross-border transmission pollution tracing

[0124] In the event of regional sand cross-border transmission, the system identifies through meteorological big data that the dominant wind direction is northwest wind, and the average wind speed is 8.4m / s. The dynamic algorithm selection module detects that the monitoring network points are arranged in an arc shape with a span of 200 kilometers, automatically switches to the arc point distribution optimization scheme and activates the aerosol ultraviolet scattering correction algorithm.

[0125] The mixed optimization module first performs coarse search with a grid size of hundreds of kilometers and a resolution of 10km x 10km, and locks three potential sand source areas after 800 genetic iterations: Mongolian Gobi area, border coal mine belt, and local construction site; Then start three-stage refined tracing: in the first stage, search the Mongolian Gobi area, step size δ0=5km, and through 12 times of axial movement, the coordinates are reduced to north latitude 42.36°+1.2km; In the second stage, the adaptive mutation strategy is used for the border coal mine belt, and the mutation rate is increased from 5% to 15%, breaking the fitness platform caused by terrain shielding; In the third stage, near-field DNA comparison is started for the local source, and the historical pollution source gene library is called to match the dust characteristic spectrum of the construction site.

[0126] Finally, the multi-source analysis module outputs the responsibility allocation with a confidence level of 95%: cross-border transmission contribution 91.2%, Mongolian source intensity [4.8 x 106 t, 5.1 x 10 6 t], local contribution 8.8%, site source [0.4 x 10 6 t, 0.45 x 10 6 t].

[0127] Example Four, Real-time Tracking of Mobile Source Pollution

[0128] To track the sudden leakage event of a hazardous chemical transport vehicle, the system integrates road network monitoring and unmanned aerial vehicle mobile monitoring. The dynamic algorithm selection module selects the dynamic wind field reconstruction algorithm and loads the road underlay roughness parameter z0=0.8m according to the characteristics that the wind speed fluctuation is greater than 3 levels and the pollution group presents a strip diffusion.

[0129] The hybrid optimization module establishes a double-process cooperative mechanism: the main process inverses the leakage intensity and moving speed through genetic algorithm, and the initial population contains the combination of load weight 15t, 25t and vehicle speed 40km / h, 80km / h; the slave process corrects the vehicle position in real time through pattern search, and updates the axial detection vector after receiving new coordinates of the unmanned aerial vehicle every 5 seconds.

[0130] When the speed parameter fitness is abnormally decreased in a certain iteration, back propagation verification is triggered immediately: after finding the concentration gradient mutation at K128+300 meters, the simulated concentration field is compared with 17 road surface sensors in real time, and the vehicle speed basis is forcibly reset to 50km / h. After 8 minutes, the pollution source is locked as the propylene tank truck with license plate XYZ, the leakage intensity interval is [38kg / min, 42kg / min], and the position error is 12.3 meters. The verification module calls the GPS trajectory of the vehicle and the loading and unloading records to confirm the matching degree of 100% during the leakage period.

[0131] Example Five, Accurate Traceability of Malodorous Pollution Source

[0132] A high-intensity malodorous complaint event suddenly occurred around a municipal sewage treatment plant. The system captures the atmospheric stability jumping from D level to F level in 2 hours through the dynamic algorithm selection module, and the wind speed drops below 0.8m / s. Based on the amine characteristic spectrum of malodorous pollutants and the low wind speed diffusion characteristics, the system automatically activates the valley inversion compensation algorithm and loads the terrain elevation data.

[0133] The hybrid optimization module divides the initial population into two groups for hydrogen sulfide and ammonia optimization: the H2S population generates initial source strength based on plant boundary monitoring concentration [5 ppm, 15 ppm], and the NH3 population sets the search range based on sludge treatment area coordinates. When the genetic algorithm iterates to 900 generations, the pattern search unit detects that the NH3 objective function has a "pseudo-peak trap" - a false high fitness value is formed near the sludge dewatering workshop coordinates, triggering the three-dimensional axial detection mechanism immediately: a positive and negative bidirectional search is performed along the vertical direction with a step size of δz = 2 m, and it is found that the fitness at an elevation of +8 m actually increases by 12.7%, thus determining that the leakage point is located at the roof exhaust valve rather than the ground equipment.

[0134] The multi-source analysis module uses the odor activity value weighting cutting technique to separate the main source of malodor as the sludge anaerobic digestion tank, with a H2S contribution of 73%, and the secondary source as the compost workshop, with a NH3 contribution of 21%. Based on the results, the environmental protection department implements a transformation of the digestion tank sealing system, and 48 hours later, the malodor complaints decrease by 92%.

[0135] Example Six, Radionuclide Leakage Tracing

[0136] The environmental monitoring network around the nuclear facility detects abnormal gamma ray spectra with a characteristic peak at 1.46 MeV. The system initiates a radioactive contamination emergency tracing, and the dynamic algorithm selection module identifies that the monitoring points are distributed in concentric circles and there is atmospheric precipitation with a rainfall intensity of 4 mm / h. The wet deposition correction algorithm is automatically matched and the radionuclide migration model is loaded.

[0137] The hybrid optimization module faces three challenges for the first time:

[0138] 1. The half-life of the nuclide leads to dynamic attenuation of the concentration;

[0139] 2. The radon background interference forms noise;

[0140] 3. The shielding effect of the protective wall distorts the diffusion field.

[0141] The system creates a four-dimensional optimization framework in space and time: the time dimension injects the cesium-137 half-life function, and the spatial dimension adds the shielding attenuation coefficient of the concrete wall μ = 0.2 cm -1 . The genetic algorithm excludes nuclides with a half-life of less than one hour in the population initialization stage, and after 600 generations of evolution, it locks in two potential leakage points - the spent fuel pool at coordinate A and the maintenance access at coordinate B. At this time, the pattern search unit detects that the gradient of coordinate B is abnormally flat, and the quantum tunneling detection strategy is started: virtual observation points are set on both sides of the shielding wall for penetration calculation, and finally coordinate A is confirmed as the true leakage point, with a fitness improvement of 8.3 times.

[0142] The multi-source analysis module uses Bayesian Monte Carlo stratification to output the cesium-137 intensity interval of the spent fuel pool leakage [3.2 × 10 4Bq, 3.5 x 10 4 Bq] with a confidence probability of 99.7%, which is an error of only 0.75 meters from the component breakage location documented in the incident log.

[0143] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and illustrative figures, it should be apparent that the scope of the present application is not limited to these specific embodiments.

[0144] While the embodiments of the application have been shown and described, it is to be understood that the embodiments described are only by way of example and are not limiting of the scope of the application. Numerous other embodiments that are included within the spirit and scope of the application can be apparent to those skilled in the art from this detailed description.

Claims

1. A big data based fast atmospheric pollutant tracing system, characterized in that: Comprising: a diffusion simulation module, which constructs a Gaussian diffusion model algorithm to calculate the concentration field distribution of pollutants according to input meteorological parameters and underlying surface characteristic parameters; a hybrid optimization module, which is coupled with the diffusion simulation module and uses a genetic-pattern search hybrid algorithm to iteratively optimize pollution source parameters, including: a parameter encoding unit, which converts pollution source parameters into genotypic individuals through decimal encoding; a population evolution unit, which updates the source parameter population through selection, crossover, and mutation operations, and selects high fitness individuals based on a fitness function; a pattern search unit, which performs axial and pattern movement searches on the local optimal solution output by the genetic algorithm to achieve global optimization; a dynamic algorithm selection module, which matches the optimal source tracing algorithm type from a preset algorithm library according to meteorological data characteristics and observation point schemes; a multi-source analysis module, which processes multiple point source scenarios in parallel to output the emission intensity confidence interval of each pollution source; a verification and output module, which verifies the model accuracy based on simulation data and Prairie Grass field test data, and outputs the source tracing results and convergence indicators.

2. The big data based atmospheric pollutant rapid tracing system according to claim 1, characterized in that: The diffusion simulation module further comprises: reliability verification of the Gaussian diffusion model based on standardized mean error, standardized mean deviation, and root mean square error indicators; calibration of model parameters through measured concentration data from the Prairie Grass project to ensure the simulation accuracy of high-value and distribution areas of pollutants.

3. The big data based atmospheric pollutant rapid tracing system according to claim 1, characterized in that: The fitness function of the hybrid optimization module is defined as: where C 模拟,i is the simulated concentration value at the nth observation point, C 观测,i is the measured concentration value at the nth observation point, and n is the total number of observation points.

4. The big data based atmospheric pollutant rapid tracing system according to claim 1, characterized in that: The operations of the parameter encoding unit include: construction of a key emission source information library, random extraction of individuals to generate an initial population; comparison of simulated values and observed values through parallel computing to filter individuals with fitness higher than a threshold to enter the next generation evolution.

5. The big data based atmospheric pollutant rapid tracing system according to claim 1, characterized in that: The dynamic algorithm selection module matches algorithms according to the following rules: when the observation point scheme is an arc scheme, preferentially select the near-source sparse point strategy; when the observation point scheme is a rectangular grid scheme, preferentially select the near-source dense point strategy; when the atmospheric stability is unstable, increase the weight of observation points beyond 400 meters; when the atmospheric stability is stable, increase the weight of observation points within 200 meters.

6. The big data based atmospheric pollutant rapid tracing system according to claim 1, characterized in that: The verification process of the verification and output module includes: simulation evaluation: generate ideal concentration fields based on Prairie Grass meteorological conditions to test the source tracing success rate and timeliness under different point schemes, point densities, and stability conditions; field verification: compare the coordinate error distance and intensity deviation of the source tracing results and the true source parameters to quantify the model uncertainty.

7. The big data based atmospheric pollutant rapid tracing system according to claim 1, characterized in that: The method comprises the following steps: S1. Input meteorological data and observed pollutant concentration data; S2. Calculate the concentration field distribution of each source parameter individual through the diffusion simulation module; S3. Calculate the fitness value of the population individuals based on the error between observed concentrations and simulated concentrations; S4. Perform genetic algorithm operations: selection operation: retain the top 30% of individuals in terms of fitness; crossover operation: exchange genes between pairs of remaining individuals at a ratio of 60%; mutation operation: randomly mutate individual gene values at a probability of 5%; S5. When the genetic algorithm iteration reaches 1000 times, input the current optimal individual into the pattern search algorithm: set the initial step size δ0=0.1 and the contraction factor α=0.5; The search range is dynamically reduced by the step update formula: δ k+1 = a - δ k where δ k is the search step size for the kth iteration, a is the step size decay factor, and k is the iteration number index; S6. When the error is less than 0.01 and the iteration reaches 2000 times, output the global optimal source parameter (Q, x, y).

8. The big data based atmospheric pollutant rapid tracing system according to claim 7, characterized in that: Before S1 step, also includes: Dynamic adjustment of observation point strategy: If the atmospheric stability is C level, 50% of the observation points are arranged in the area 400 meters away from the source; If the atmospheric stability is E level, 70% of the observation points are arranged in the area within 200 meters from the source. 9.The big data based atmospheric pollutant rapid tracing system according to claim 7, characterized in that: The specific implementation of the cross operation in S4 is: Single-point crossover method is adopted, a cutting point in the gene sequence is randomly selected, and the gene fragments after cutting of the parent individuals are exchanged; After generating the offspring individuals, the fitness is recalculated and the individuals below the average value are eliminated.

10. The big data based fast atmospheric pollutant tracing system according to claim 7, wherein: The termination condition of the mode search in S5 is: Step size delta k Less than 0.001 and no better solution found for 50 consecutive iterations Before outputting the final solution, the correlation coefficient with the observed concentration is calculated, and if it is lower than 0.95, the genetic algorithm restart mechanism is triggered.

Images