An air pollution source inversion method based on improved ant colony algorithm

Abstract

The present invention discloses an air pollution source inversion method based on an improved ant colony algorithm. The method mainly comprises: establishing a gas diffusion model according to the gas diffusion law, and establishing an air pollution source inversion model based on this model; secondly, solving the model by using an improved ant colony algorithm. The improved ant colony algorithm is improved in two aspects: the ant colony algorithm is prone to falling into local extreme values and the convergence speed. The crossover idea of the genetic algorithm and the pheromone update mechanism of the ant colony algorithm are combined to enhance the global search capability of the ant colony algorithm and accelerate the convergence speed, thereby achieving rapid and accurate inversion of relevant parameter information of air pollution sources, achieving rapid positioning of pollution sources, and carrying out targeted treatment. At the same time, the air pollution source inversion method based on the improved ant colony algorithm proposed in the present invention can also be applied to source inversion problems such as water pollution and soil pollution.

Description

Technical Field

[0001] The invention discloses a pollution source inversion method based on a swarm intelligence algorithm, in particular to an air pollution source inversion method based on an improved ant colony algorithm. Background Art

[0002] With the development of the chemical industry, air pollution has become a problem that cannot be ignored. Only by clearly identifying the location and emission volume of pollution sources can we lay a solid foundation for further pollution control and enable the entire air monitoring and governance chain to achieve precise monitoring, real-time source analysis, timely management, and targeted governance.

[0003] The traditional method for locating air pollution sources is mainly to check through enterprise emission source lists. However, due to the open nature of air pollutants in the monitoring area, obtaining the initial emission list is too difficult and the data is incomplete. At the same time, the existence of some unknown enterprises has made it impossible to collect a complete source list. In addition, due to the incompleteness of activity level data, emission factors are difficult to determine. However, even if a complete database can be established and emission data for all pollution sources in the monitoring area can be obtained, the rapid development of my country's industry will make the data volume of the entire pollution source list too large, making it very difficult and inefficient to check it through manual labor alone. Summary of the Invention

[0004] In order to improve the positioning accuracy and investigation efficiency of air pollution sources and meet the needs of precise positioning and targeted treatment of pollution sources, the present invention provides an air pollution source inversion method based on an improved ant colony algorithm, which can effectively locate and invert the parameters of pollution sources that cause air pollution exceeding the standard, and can improve the efficiency and accuracy of existing methods.

[0005] The global optimization search method uses artificial intelligence optimization methods. Based on the characteristics of gas diffusion, wind speed, wind direction, and other factors, an appropriate gas diffusion model is selected. A loss function is established as the model. The loss function is repeatedly solved using an artificial intelligence algorithm to obtain the location and emission volume of air pollution emission sources, which can reduce labor costs and improve efficiency. Therefore, the research on inverting air pollution sources using swarm intelligence algorithms has certain theoretical significance and application value.

[0006] The technical solution adopted by the present invention to achieve the above-mentioned purpose is: an air pollution source inversion method based on an improved ant colony algorithm, comprising the following steps:

[0007] Step 1: Collect monitoring data from all monitoring stations where air pollution exceeds the standard;

[0008] Step 2: Construct and optimize the gas diffusion model, which together with the target optimization function constitutes the air pollution source inversion model;

[0009] Step 3: Use the ant colony algorithm to iteratively solve the air pollution source inversion model to obtain pollution source information to determine the air pollution source.

[0010] Step 2 is as follows:

[0011] Build a gas diffusion model:

[0012]

[0013] Where C(x,y,z) represents the simulated gas concentration at any point (x,y,z) downwind; Q0 represents the intensity of the emission source; u represents the average wind speed; x0 and y0 represent the horizontal coordinates of the emission source; x and y represent the horizontal coordinates of the monitoring station; H0 represents the height of the emission source; z represents the vertical coordinate of any point (x,y,z) downwind of the emission outlet; σ y and σ z represents the horizontal and vertical diffusion coefficients, expressed as:

[0014]

[0015]

[0016] Where ω1, ω2, θ1, θ2 are coefficients;

[0017] The gas diffusion model is optimized as follows: Let z = 0, then the optimized gas diffusion model is as follows:

[0018]

[0019] The standard loss function is constructed based on the optimized gas diffusion model as follows:

[0020]

[0021] in, Indicates the actual pollutant concentration detected by the monitoring station. represents the pollutant concentration obtained by the above optimized gas diffusion model, L tar Represents the loss function value, which is used to evaluate the accuracy of the current result. The closer it is to 0, the more accurate the current result is. T represents the number of monitoring stations.

[0022] Step 3 is as follows:

[0023] Step 3.1: Initialize various parameters: the importance factors α and β of the pheromone and heuristic function, the ant colony size m, the number of iterations N, the pheromone volatility factor ρ, and the upper and lower limits of the emission source parameters, including the emission source intensity Q0, location (x0, y0), and release height H0;

[0024] Step 3.2: Randomly generate ant individuals within the feasible domain of various source parameters and initialize the pheromone τ of each ant i (i=1,2,3,…,m);

[0025] Step 3.3: Each ant chooses its next position according to the state transition rule and search strategy;

[0026] Step 3.4: After all ants have completed their search, update the current global optimal solution and then perform selection and exchange operations;

[0027] Step 3.5: Update the ant individuals according to the relationship between the loss value of the ith ant in the n+1 generation and the loss value of its nth generation; when the number of iterations reaches N, if the loss function L tar If convergence occurs, the ant individual with the current iteration number is taken as the optimal ant individual, and the pollution source is checked based on the corresponding source parameters.

[0028] The step 3.3 is achieved by the following steps:

[0029] Step 3.3.1: The state transition probability is as follows:

[0030]

[0031] Where, represents the heuristic factor P i (n) represents the probability of the i-th ant choosing the next position, when P i The larger (n) is, the closer the current solution is to the source parameters of the pollution source; τ i represents the pheromone of the i-th ant, η i represents the heuristic function of the i-th ant; n is the current number of iterations;

[0032] Step 3.3.2: Set a state transition constant P0. When P i When (n) is greater than P0, a local search strategy is adopted, otherwise a global search strategy is adopted, which can be expressed as follows:

[0033]

[0034] Among them, next is the next position of the current ant; now is the current position of the current ant; rand is a random number between (0,1); step is the local search step; λ is the step coefficient, which is the inverse of the number of iterations. Range represents the feasible domain range.

[0035] The step 3.4 is as follows:

[0036] After each iteration, the entire ant colony is sorted from small to large according to the loss value, and the one with the largest loss value is ranked first. Individuals are discarded, and the remaining Randomly select individuals with loss values in the middle position to replicate and fill in the discarded thus forming new populations;

[0037] Randomly select two individuals P1 and P2 from the new population for crossover, and exchange parts of P1 and P2 to form new individuals. The crossover method is as follows:

[0038]

[0039] Among them, P1 and P2 represent the selected ant individuals, and the vectors composed of the source parameters of the emission sources included. Indicates that a new ant individual is formed after crossing, A and B represent the crossing points, ranging from (0,1) and A<B.

[0040] The updating of ant individuals is achieved by the following formula:

[0041] τ i (n+1)=(1-ρ)τ i (n)+Δτ i (n)

[0042]

[0043] Where n is the current iteration number; ρ represents the pheromone volatility factor; τ i represents the pheromone of the i-th ant; Δτ i represents the pheromone increment of the i-th ant; K represents the constant coefficient; L i Represents the loss value of the i-th ant individual; μ represents a coefficient. For the i-th ant, if its loss value of the n-1 generation is greater than the loss value of the n-th generation, then its pheromone increment Δτ i Increase, otherwise decrease.

[0044] The ant individual with the current number of iterations is used as the optimal ant individual, and the pollution source is checked based on the corresponding source parameters, specifically:

[0045] Taking the coordinates (x, y) of the optimal ant individual as the center and setting the length R as the radius, within this circular area, the emission sources of a certain type of pollutants are ranked according to their distance from the inverted pollution source, and the predicted pollution contribution rate of each emission source is obtained to determine the air pollution source.

[0046] The contribution rate formula is as follows:

[0047]

[0048] Among them, P i represents the pollution probability of the i-th emission source; D i Represents the distance from the i-th emission source to the plane coordinate (x, y) corresponding to the optimal solution; M is the number of emission sources within the investigation range.

[0049] An air pollution source inversion system based on an improved ant colony algorithm includes:

[0050] A data acquisition module is used to obtain monitoring data from all monitoring stations where air pollution exceeds the standard;

[0051] The model building module is used to build and optimize the gas diffusion model, which together with the target optimization function constitutes the air pollution source inversion model;

[0052] The inversion module is used to iteratively solve the air pollution source inversion model through the ant colony algorithm to obtain pollution source information to determine the air pollution source.

[0053] The beneficial effects and advantages of the present invention are:

[0054] The present invention uses an improved ant colony algorithm and utilizes the crossover operation of the genetic algorithm to solve the shortcomings of the traditional ant colony algorithm that is prone to falling into local optimality, improves the pheromone update mechanism, solves the slow convergence speed of the traditional ant colony algorithm, further enhances the search capability, and can more quickly and accurately invert the source parameter information of pollutants that cause air pollution alarms. The effectiveness of the method has been verified through relevant experiments, providing practical significance for air pollution control.

[0055] A gas diffusion model is established based on the gas diffusion law, and an air pollution source inversion model is established based on this. Secondly, for solving the model, an improved ant colony algorithm is used to solve it. The ant colony algorithm is improved in two aspects: it is easy to fall into local extreme values and its convergence speed. In combination with the crossover idea of the genetic algorithm and the pheromone update mechanism of the updated ant colony algorithm, the global search capability of the ant colony algorithm is enhanced and the convergence speed is accelerated, thereby achieving rapid and accurate inversion of relevant parameter information of air pollution sources, achieving rapid positioning of pollution sources, and carrying out targeted treatment. At the same time, the air pollution source inversion method based on the improved ant colony algorithm proposed in the present invention can also be applied to source inversion problems such as water pollution and soil pollution. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a flowchart of an air pollution source inversion method based on an improved ant colony algorithm;

[0057] Figure 2 This is a flowchart of the improved ant colony algorithm. DETAILED DESCRIPTION

[0058] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the technical solutions of the present invention are further described below with reference to examples. The following description sets forth many specific details to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways than those described herein, and those skilled in the art can make similar improvements without violating the scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art of the art to which the present invention pertains. The terms used in the specification of the invention herein are for the purpose of describing specific embodiments only and are not intended to limit the present invention.

[0060] An air pollution source inversion method based on an improved ant colony algorithm comprises the following steps:

[0061] Step 1: Observe the monitoring data of the deployed monitoring stations. When a monitoring station issues an air pollution exceeding standard alert, collect and record the monitoring data of all monitoring stations at this time.

[0062] Step 2: Establish a suitable gas diffusion model based on turbulent diffusion theory, and modify the gas diffusion model according to actual needs and conditions. Then, based on the modified gas diffusion model, establish a target optimization function, which is the air pollution source inversion model.

[0063] Step 3: Improve the traditional ant colony algorithm, including correcting its shortcomings of being easily trapped in local optimal values and having slow initial convergence speed.

[0064] Step 4: Use the improved ant colony algorithm to iteratively solve the air pollution source inversion model, and finally obtain the relevant parameter information of the pollution source.

[0065] Step 5: Based on the obtained information about the pollution source, check the enterprises in a small surrounding area, and eventually identify the enterprises whose pollution emissions exceed the standard.

[0066] The monitoring data of the monitoring station is the real-time concentration of a series of pollutants monitored by the monitoring station and related meteorological parameters.

[0067] The air pollution exceeding standard alarm is an air pollution alarm issued when a certain air pollutant monitored by the monitoring station exceeds the alarm threshold.

[0068] The collecting and recording of monitoring data from all monitoring stations at this time is to record the real-time concentration of pollutants that cause the air pollution exceeding the standard alarm.

[0069] The appropriate gas diffusion model is the diffusion model of an elevated continuous point source, which is expressed as follows:

[0070]

[0071] Where C(x, y, z) represents the simulated gas concentration at any point downwind; Q0 represents the intensity of the emission source; u represents the average wind speed; x0 and y0 represent the plane coordinates of the emission source; x, y and represent the plane coordinates of the monitoring station; H0 represents the height of the emission source; z represents the vertical coordinate of any point downwind of the emission outlet; σ y and σ z The diffusion coefficient representing the vertical and vertical downwind directions can be determined according to the diffusion coefficient equation of the GIFFORD model, and its general expression is:

[0072]

[0073] Where ω1, ω1, θ1, and θ2 are coefficients whose values are determined by the current atmospheric stability level; x and x0 are the horizontal coordinates of any point downwind of the pollution source and the horizontal coordinates of the location of the pollution source, respectively.

[0074] The modified gas diffusion model only needs to focus on the distribution of pollutant concentration near the ground. Let z = 0, and its expression is as follows:

[0075]

[0076] The air pollution source inversion model is a target loss function established based on the modified gas diffusion model, using a square loss function, which is expressed as follows:

[0077]

[0078] in, Indicates the actual pollutant concentration detected by the monitoring station. represents the pollutant concentration obtained by simulating the above gas diffusion model. According to the above formula, the problem of source parameter inversion can be transformed into the problem of L tar Perform the optimization problem so that L tar The minimum solution is the relevant parameters of the pollution source that need to be inverted.

[0079] The traditional ant colony algorithm is a swarm intelligence algorithm that utilizes the pheromones left by ants during the foraging process to continuously adjust the next movement direction of the entire population, thereby iteratively finding the optimal foraging path.

[0080] The improved ant colony algorithm is an improved ant colony algorithm that addresses the shortcomings of the traditional ant colony algorithm, such as being prone to falling into local optimality and slow convergence. It includes the following steps:

[0081] Step 1: Analogously to the crossover operation in the genetic algorithm, select and exchange to construct different ants, maintain the diversity of the entire ant colony, and ensure that the algorithm can escape from local extremes.

[0082] (1) After each iteration, the entire ant colony is sorted from small to large in terms of loss, and the one with the largest loss is ranked. Individuals are discarded, and the remaining Randomly select individuals with loss degree in the middle position to replicate and fill in the discarded thus forming a new population.

[0083] (2) Randomly select two individuals P1 and P2 from the new population and perform crossover. Similar to the partial mapping crossover in the genetic algorithm, part of P1 and P2 are exchanged to form two new individuals. If the two new individuals are better than the original two individuals, they are replaced.

[0084] The specific crossover method is as follows:

[0085]

[0086] Where P1 and P2 represent the matrix composed of the parameters related to the assumed pollution sources contained in the selected ants. Indicates that a new ant individual is formed after crossing, A and B represent the crossing points, ranging from (0,1) and A<B.

[0087] Step 2: Improve the pheromone update mechanism to speed up the algorithm's search efficiency. The relationship between the loss value of the ith ant in the n+1 generation and the loss value of its nth generation determines whether to increase or decrease the pheromone of the current feasible solution. The entire update process is as follows:

[0088] τ i (n+1)=(1-ρ)τ i (n)+Δτ i (n)

[0089]

[0090] Where n is the current iteration number; ρ represents the pheromone volatility factor; τi represents the pheromone of the i-th ant; Δτ i represents the pheromone increment of the i-th ant; K represents the constant coefficient; L i represents the loss value of the i-th ant; μ represents a coefficient. For the i-th ant, if its loss value of the n-1 generation is greater than the loss value of the n-1 generation, then its pheromone increment will be increased, otherwise it will be reduced.

[0091] The improved ant colony algorithm air pollution source inversion model is iteratively solved, including the following steps:

[0092] Step 1: Collect the measured concentrations of the main pollutants that cause air pollution alerts The location information (x, y) corresponding to the i-th monitoring station is substituted into the target loss function L tar middle.

[0093] Step 2: Determine the current atmospheric stability level based on the relevant meteorological parameters of the monitoring station, and determine σ by referring to the diffusion coefficient equation of the GIFFORD model and the current atmospheric stability level. y and σ z The value of is substituted into the target loss function L tar middle.

[0094] Step 3: Set the parameters required by the algorithm, including the importance factors α and β of the information and heuristic functions, as well as the ant colony size m, the number of iterations N, the pheromone volatility factor ρ, the pheromone matrix X, and the upper and lower limits of each parameter.

[0095] Step 4: Start the algorithm and iteratively solve the target loss function until the Nth iteration, and record the target loss value and the corresponding optimal solution at this time.

[0096] The pollution source enterprise screening is to screen the enterprises related to the relevant pollutant emissions in an area with R as the radius according to the inverted relevant pollution source parameters, and give a predicted pollution contribution rate ranking of each enterprise according to the distance from the inverted pollution source.

[0097] like Figure 1 As shown in FIG, an air pollution source inversion method based on an improved ant colony algorithm, taking the PM2.5 pollution exceeding standard alarm as an example, includes the following steps:

[0098] Step 1: Data collection, recording, and preprocessing of exceeded-standard alarms. Observe the monitoring status of the entire monitoring device deployment network and set the PM2.5 sensitivity threshold to when one or several monitoring devices issue an exceeded-standard PM2.5 pollution alarm. Observe for 5 minutes, then record the real-time PM2.5 concentration data of all devices that issued exceeded-standard pollution alarms. Also record the wind direction and wind speed information for these devices. Eliminate abnormal pollutant data or meteorological data due to equipment problems (equipment damage, maintenance, etc.) or human manipulation (deletion, modification, etc.). Use the remaining data as the data set for this source inversion.

[0099] Step 2: Establish the target loss function. Using the modified ground concentration distribution model downwind of the pollution source (i.e., z = 0) based on the elevated point source continuous diffusion model (i.e., Gaussian plume model), establish the target loss function (e.g., squared loss function, mean square error function, R 2 Loss function, etc.), here we take the square loss function as an example:

[0100]

[0101] Among them L tar is the target loss value, and are the actual concentration and predicted concentration of air pollutants corresponding to the i-th detection device respectively. tar When it approaches 0, it indicates that the current solution is closer to the actual pollution source.

[0102] Step 3: Determine σ by referring to the diffusion coefficient equation of the GIFFORD model y and σ z The value of is determined by the real-time atmospheric stability and is generally expressed as:

[0103]

[0104] Where x and x0 are the horizontal coordinates of any point downwind from the pollution source and the location of the pollution source, respectively; ω1, ω2, θ1, and θ2 are coefficients whose values are determined by the current atmospheric stability level.

[0105] Step 4: Use M-ACO (improved ant colony algorithm) to iteratively solve. Figure 2 As shown, the following steps are included:

[0106] Step 4.1: Initialize the algorithm parameters, the importance factors α and β of pheromone and heuristic function, the ant colony size m, the number of iterations N, the pheromone volatilization factor ρ, and the upper and lower limits of the source parameters (source strength Q, location x0, y0, and release height H).

[0107] Step 4.2: Randomly generate ant individuals within the feasible domain formed by various source parameters and initialize the pheromone concentration of each ant.

[0108] Step 4.3: Each ant chooses its next move based on the state transition rules and search strategy.

[0109] Step 4.3.1: The state transition probability is as follows:

[0110]

[0111] Among them, P i (n) represents the probability of the i-th ant choosing the next position, when P i The larger (n) is, the greater the probability that the ant individual will be selected next time.

[0112] Step 4.3.2: Set a state transition constant P0. When P i When (n) is greater than P0, a local search strategy is adopted, otherwise a global search strategy is adopted, which can be expressed as follows:

[0113]

[0114] Where next is the next position of the current ant; now is the current position of the current ant; rand is a random number between (0,1); step is the local search step; λ is the step coefficient, which is the inverse of the number of iterations. Range represents the feasible domain range.

[0115] Step 4.4: When all ants have finished searching, update the current global optimal solution, and then perform selection and exchange operations. After each iteration, sort the entire ant colony from small to large loss, and sort the ant colony with the largest loss. Individuals are discarded, and the remaining Randomly select individuals with loss degree in the middle position to replicate and fill in the discarded thus forming a new population.

[0116] Randomly select two individuals P1 and P2 from the new population and perform crossover. Similar to the partial mapping crossover in the genetic algorithm, part of P1 and P2 are exchanged to form a new individual. The specific crossover method is as follows:

[0117]

[0118] Where P1 and P2 represent the matrix composed of the parameters related to the assumed pollution sources contained in the selected ants. Indicates that a new ant individual is formed after crossing, A and B represent the crossing points, ranging from (0,1) and A<B.

[0119] Step 4.5: Based on the relationship between the loss value of the ith ant in the n+1 generation and the loss value of its nth generation, decide whether to increase or decrease the pheromone of the current feasible solution. The entire update process is as follows:

[0120] τ i (n+1)=(1-ρ)τ i (n)+Δτ i (n)

[0121]

[0122] Where n is the current iteration number; ρ represents the pheromone volatility factor; τ i represents the pheromone concentration of the i-th ant; Δτ i represents the pheromone increment of the i-th ant; K represents the constant coefficient; L i represents the loss value of the i-th ant; μ represents a coefficient. For the i-th ant, if its loss value of the n-1 generation is greater than the loss value of the n-1 generation, then its pheromone increment will be increased, otherwise it will be reduced.

[0123] Step 5: After the number of iterations in step 4 reaches N, observe the loss function L tar Convergence. If converged, the pollution source is checked based on the pollution source parameters corresponding to the optimal ant individual. With the coordinates (x, y) of the optimal solution as the center and a certain length R as the radius, within this circular area, the relevant enterprises involved in PM2.5 emissions are listed and a pollution probability check list for each enterprise is given. The probability calculation formula is as follows:

[0124]

[0125] Among them, P i represents the pollution probability of the i-th enterprise; D i Represents the distance from the i-th enterprise to the plane coordinate (x, y) corresponding to the optimal solution; M is the number of related enterprises within the investigation range

[0126] By checking this probability list, we can find the company that caused the PM2.5 pollution accident. At the same time, we can also compare the results of the investigation with the inversion results to evaluate the accuracy of the inversion.

[0127] The above-mentioned specific implementation methods are used to illustrate the present invention and are only preferred embodiments of the present invention, rather than limiting the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit of the present invention and the scope of protection of the claims shall fall within the scope of protection of the present invention.

Claims

1. An air pollution source inversion method based on an improved ant colony algorithm, characterized in that: The following steps are involved: Step 1: Collect monitoring data from all monitoring stations where air pollution exceeds the standard; Step 2: Construct and optimize the gas diffusion model, which together with the target optimization function constitutes the air pollution source inversion model; Step 3: Use the ant colony algorithm to iteratively solve the air pollution source inversion model to obtain pollution source information to determine the air pollution source; Step 3 is as follows: Step 3.1: Initialize various parameters: the importance factors α and β of the pheromone and heuristic function, the ant colony size m, the number of iterations N, the pheromone volatility factor ρ, and the upper and lower limits of the emission source parameters, including the emission source intensity Q0, location (x0, y0), and release height H0; Step 3.2: Randomly generate ant individuals within the feasible domain of various source parameters and initialize the pheromone τ of each ant i , where i = 1, 2, 3, …, m; Step 3.3: Each ant chooses its next position according to the state transition rule and search strategy; Step 3.4: After all ants have completed their search, update the current global optimal solution, and then perform selection and exchange operations; Step 3.5: Update the ant individuals according to the relationship between the loss value of the ith ant in the n+1 generation and the loss value of its nth generation; when the number of iterations reaches N, if the loss function L tar If convergence occurs, the ant individual with the current number of iterations is used as the optimal ant individual, and the pollution source is checked based on the corresponding source parameters; The step 3.4 is as follows: After each iteration, the entire ant colony is sorted from small to large according to the loss value, and the one with the largest loss value is ranked first. Individuals are discarded, and the remaining Randomly select individuals with loss values in the middle position to replicate and fill in the discarded thus forming new populations; Randomly select two individuals P1 and P2 from the new population for crossover, and exchange parts of P1 and P2 to form new individuals. The crossover method is as follows: Among them, P1 and P2 represent the selected ant individuals, and the vectors composed of the source parameters of the emission sources included. Indicates the formation of a new ant individual after crossing, A and B represent the crossing points, ranging from (0,1) and A<B; The updating of ant individuals is achieved by the following formula: t i (n+1)=(1-ρ)τ i (n)+Δτ i (n) Where n is the current iteration number; ρ represents the pheromone volatility factor; τ i represents the pheromone of the i-th ant; Δτ i represents the pheromone increment of the i-th ant; K represents the constant coefficient; L i Represents the loss value of the i-th ant individual; μ represents a coefficient. For the i-th ant, if its loss value of the n-1 generation is greater than the loss value of the n-th generation, then its pheromone increment Δτ i Increase, otherwise decrease.

2. The air pollution source inversion method based on the improved ant colony algorithm according to claim 1 is characterized in that: Step 2 is as follows: Build a gas diffusion model: Where C(x,y,z) represents the simulated gas concentration at any point (x,y,z) downwind; Q0 represents the intensity of the emission source; u represents the average wind speed; x0 and y0 represent the horizontal coordinates of the emission source; x and y represent the horizontal coordinates of the monitoring station; H0 represents the height of the emission source; z represents the vertical coordinate of any point (x,y,z) downwind of the emission outlet; σ y and σ z represents the horizontal and vertical diffusion coefficients, expressed as: Where ω1, ω2, θ1, θ2 are coefficients; The gas diffusion model is optimized as follows: Let z = 0, then the optimized gas diffusion model is as follows: The standard loss function is constructed based on the optimized gas diffusion model as follows: in, Indicates the actual pollutant concentration detected by the monitoring station. represents the pollutant concentration obtained by the above optimized gas diffusion model, L tar Represents the loss function value, which is used to evaluate the accuracy of the current result. The closer it is to 0, the more accurate the current result is. T represents the number of monitoring stations.

3. The air pollution source inversion method based on the improved ant colony algorithm according to claim 1 is characterized in that: The step 3.3 is achieved by the following steps: Step 3.3.1: The state transition probability is as follows: Where, represents the heuristic factor P i (n) represents the probability of the i-th ant choosing the next position, when P i The larger (n) is, the closer the current solution is to the source parameters of the pollution source; τ i represents the pheromone of the i-th ant, η i represents the heuristic function of the i-th ant; n is the current number of iterations; Step 3.3.2: Set a state transition constant P0. When P i When (n) is greater than P0, a local search strategy is adopted, otherwise a global search strategy is adopted, which can be expressed as follows: Among them, next is the next position of the current ant; now is the current position of the current ant; rand is a random number between (0,1); step is the local search step; λ is the step coefficient, which is the inverse of the number of iterations. Range represents the feasible domain range.

4. The air pollution source inversion method based on the improved ant colony algorithm according to claim 1 is characterized in that: The ant individual with the current number of iterations is used as the optimal ant individual, and the pollution source is checked based on the corresponding source parameters, specifically: Taking the coordinates (x, y) of the optimal ant individual as the center and setting the length R as the radius, within this circular area, the emission sources of a certain type of pollutants are ranked according to their distance from the inverted pollution source, and the predicted pollution contribution rate of each emission source is obtained to determine the air pollution source.

5. The air pollution source inversion method based on the improved ant colony algorithm according to claim 4 is characterized in that: The contribution rate formula is as follows: Among them, P i represents the pollution probability of the i-th emission source; D i Represents the distance from the i-th emission source to the plane coordinate (x, y) corresponding to the optimal solution; M is the number of emission sources within the investigation range.

6. An air pollution source inversion system based on an improved ant colony algorithm, characterized in that: include: A data acquisition module is used to obtain monitoring data from all monitoring stations where air pollution exceeds the standard; The model building module is used to build and optimize the gas diffusion model, which together with the target optimization function constitutes the air pollution source inversion model; The inversion module is used to iteratively solve the air pollution source inversion model through the ant colony algorithm to obtain pollution source information and determine the air pollution source; The inversion module is configured to perform: Step 3.1: Initialize various parameters: the importance factors α and β of the pheromone and heuristic function, the ant colony size m, the number of iterations N, the pheromone volatility factor ρ, and the upper and lower limits of the emission source parameters, including the emission source intensity Q0, location (x0, y0), and release height H0; Step 3.2: Randomly generate ant individuals within the feasible domain of various source parameters and initialize the pheromone τ of each ant i , where i = 1, 2, 3, …, m; Step 3.3: Each ant chooses its next position according to the state transition rule and search strategy; Step 3.4: After all ants have completed their search, update the current global optimal solution, and then perform selection and exchange operations; Step 3.5: Update the ant individuals according to the relationship between the loss value of the ith ant in the n+1 generation and the loss value of its nth generation; when the number of iterations reaches N, if the loss function L tar If convergence occurs, the ant individual with the current number of iterations is used as the optimal ant individual, and the pollution source is checked based on the corresponding source parameters; The step 3.4 is as follows: After each iteration, the entire ant colony is sorted from small to large according to the loss value, and the one with the largest loss value is ranked first. Individuals are discarded, and the remaining Randomly select individuals with loss values in the middle position to replicate and fill in the discarded thus forming new populations; Randomly select two individuals P1 and P2 from the new population for crossover, and exchange parts of P1 and P2 to form new individuals. The crossover method is as follows: Among them, P1 and P2 represent the selected ant individuals, and the vectors composed of the source parameters of the emission sources included. Indicates the formation of a new ant individual after crossing, A and B represent the crossing points, ranging from (0,1) and A<B; The updating of ant individuals is achieved by the following formula: t i (n+1)=(1-ρ)τ i (n)+Δτ i (n) Where n is the current iteration number; ρ represents the pheromone volatility factor; τ i represents the pheromone of the i-th ant; Δτ i represents the pheromone increment of the i-th ant; K represents the constant coefficient; L i Represents the loss value of the i-th ant individual; μ represents a coefficient. For the i-th ant, if its loss value of the n-1 generation is greater than the loss value of the n-th generation, then its pheromone increment Δτ i Increase, otherwise decrease.

Images