Multi-objective optimization method and system for water pollution load distribution

Abstract

The present invention provides a multi-objective optimization method and system for water pollution load distribution, which relates to the technical field of water pollution load distribution. The present invention divides a target water body into multiple pollution areas and obtains historical pollution discharge data of sewage outlets; constructs a water pollutant concentration prediction model and a treatment cost prediction model through a deep neural network, respectively, for predicting water pollution concentration and treatment cost; calculates the actual pollutant discharge amount, pollutant reduction amount and treatment load of the sewage outlet based on the pollutant discharge allocation amount, sewage discharge flow rate, etc.; finally, with the goals of minimizing the total treatment cost, minimizing the water pollutant concentration and minimizing the pollutant treatment load, combined with constraints, determines the optimal distribution plan of pollutant discharge through a genetic algorithm, thereby achieving efficient optimization of water pollution control.

Description

Technical Field

[0001] The present invention relates to the technical field of water pollution load distribution, and in particular to a multi-objective optimization method and system for water pollution load distribution. Background Art

[0002] With the rapid development of society and the economy, water pollution control has become a crucial task for achieving sustainable development and safeguarding ecological and environmental safety. However, due to the widespread distribution of water pollution sources, the significant differences in pollution levels across regions, and the significant differences in control capabilities and costs between different pollution sources, traditional water pollution control methods often struggle to effectively reduce pollutant concentrations while balancing control costs and resource allocation fairness. Therefore, achieving a reasonable trade-off between pollutant concentration reduction, control costs, and control fairness in water pollution control has become a major technical challenge in current research on water pollution load allocation.

[0003] Current research focuses on single-objective optimization, such as reducing pollutant concentrations in a specific area, while neglecting cost-effectiveness or fairness across different regions. Other approaches prioritize cost control over water quality improvements. Furthermore, traditional optimization algorithms have limitations in efficiency and accuracy for complex multi-objective problems. Therefore, a multi-objective optimization approach to water pollution load allocation is urgently needed that can simultaneously achieve the goals of reducing pollutant concentrations, controlling treatment costs, and ensuring equitable allocation of treatment resources. This approach, coupled with effective optimization algorithms to enhance computational efficiency and decision-making quality, can also be employed.

[0004] In the prior art, publication number CN110619435A discloses a two-layer multi-objective optimization method for water pollution load distribution, which determines the environmental Gini coefficient based on the environmental Gini coefficient index and the weights of each environmental Gini coefficient index; performs upper-layer load distribution based on a two-layer multi-objective optimization water pollution load distribution model based on the minimum environmental Gini coefficient and the minimum unit pollutant emission cost; and performs lower-layer load distribution based on a two-layer multi-objective optimization water pollution load distribution model based on the maximum total industrial output value and the minimum unevenness of the reduction rate of each sewage outlet.

[0005] The main problems with the above method are: the two-layer optimization method divides the objectives into two layers for processing, the upper layer focuses on the environmental Gini coefficient and unit pollution discharge cost, and the lower layer focuses on the gross industrial output value and uneven reduction rate, which to a certain extent leads to conflicts or compromises between the upper and lower layer objectives, thereby affecting the overall optimization results; governance fairness is mainly reflected by the environmental Gini coefficient, while the environmental carrying capacity, economic development level and pollution control capacity of different regions are difficult to fully reflect through the Gini coefficient, resulting in limited applicability of the actual allocation plan.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide a multi-objective optimization method and system for water pollution load distribution to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A multi-objective optimization method for water pollution load distribution includes the following steps:

[0010] Step 1: Divide the target water body into multiple pollution zones, and obtain the pollutant discharge distribution, treatment cost, discharge flow rate, and pollutant discharge concentration of each sewage outlet during the previous discharge period, as well as the maximum water pollutant concentration of each pollution zone during the corresponding discharge period;

[0011] Step 2: Build a water pollutant concentration prediction model based on a deep neural network. The discharge flow and pollutant discharge concentration of all sewage outlets within the same discharge time period are used as input, and the maximum water pollutant concentration in the polluted area is used as a label to train the water pollutant concentration prediction model.

[0012] Step 3: Based on the pollutant emission allocation, discharge flow rate, and pollutant concentration, calculate the actual pollutant emissions, pollutant reduction, and pollutant treatment load for each discharge outlet during the discharge period;

[0013] Step 4: Build a governance cost prediction model based on a deep neural network, using the pollutant reduction, discharge flow rate, and pollutant emission concentration of each sewage outlet during the discharge period as input and the governance cost as a label to train the governance cost prediction model;

[0014] Step 5: Obtain the total amount of pollutant discharge allocation in the future pollution discharge period, and generate a fitness function with the optimization objectives of minimizing the total cost of treatment, minimizing the water pollutant concentration, and minimizing the pollutant treatment load. With the maximum water pollutant concentration in each pollution area less than the pollutant concentration standard, the sum of the pollutant discharge allocation of each sewage outlet equal to the total amount of pollutant discharge allocation, and the pollutant treatment load of each sewage outlet not greater than the maximum design load as constraints, based on the genetic algorithm, the water pollutant concentration prediction model and the treatment cost prediction model, the optimal water pollution load allocation scheme is obtained. The optimal water pollution load allocation scheme is the pollutant discharge allocation, discharge flow and pollutant discharge concentration of each sewage outlet in the future pollution discharge period.

[0015] Furthermore, the pollutant emission allocation is the amount of pollutants allocated to each sewage outlet; the treatment cost represents the economic cost incurred by reducing the pollutant emission concentration at the sewage outlet during the sewage discharge time period, which is determined by collecting historical sewage treatment financial expenditure details; the sewage flow rate represents the amount of sewage containing pollutants discharged from each sewage outlet per unit time, and an ultrasonic flow meter is installed at the sewage outlet for real-time monitoring; the pollutant emission concentration represents the concentration of the pollutant with the highest proportion in the wastewater at the sewage outlet during the sewage discharge time period, and an online pollutant concentration monitoring instrument is installed at the sewage outlet to obtain real-time data, and the online monitoring instrument includes a COD monitor and an ammonia nitrogen analyzer; the maximum water pollution concentration represents the maximum concentration of pollutants in the water body during the sewage discharge time period.

[0016] Furthermore, the principles for calculating the actual pollutant emissions, pollutant reductions, and pollutant treatment loads at each sewage outlet during the discharge period are as follows:

[0017] The formula for calculating actual pollutant emissions is:

[0018] M i =Q i ·C i ·T

[0019] Among them, M i represents the amount of pollutant discharged during the discharge period of the i-th sewage outlet, i represents the index of the sewage outlet, Q i represents the discharge flow of the i-th sewage outlet, C i represents the average pollutant concentration of the ith sewage outlet during the sewage discharge period, and T represents the duration of the sewage discharge period;

[0020] The formula for calculating pollutant reductions is:

[0021] D i =F i -M i

[0022] Among them, D i represents the pollutant reduction of the i-th sewage outlet, F i represents the pollutant emission allocation of the i-th sewage outlet;

[0023] The formula for calculating pollutant treatment load is:

[0024]

[0025] Among them, L i Represents the treatment load of the i-th sewage outlet per unit time.

[0026] Furthermore, with the optimization objectives of minimizing the total cost of treatment, minimizing the concentration of water pollutants, and minimizing the pollutant treatment load, the generated fitness function is:

[0027]

[0028] Where X represents the total cost of treatment, i represents the index of the sewage outlet, and i∈[1,N], N represents the number of sewage outlets, S i represents the treatment cost of the ith sewage outlet output by the treatment cost prediction model, X max represents the upper limit of the total cost of governance, j represents the index of the polluted area, and j∈[1,H], H represents the number of polluted areas divided into water bodies, P j It represents the maximum pollutant concentration of the jth polluted area output by the water pollutant concentration prediction model, P j,std represents the expected value of pollutant concentration in the jth polluted area, L i,max represents the maximum design load of the i-th sewage outlet, w1, w2, and w3 represent weight coefficients, and w1+w2+w3=1, w2>w3>w1.

[0029] Furthermore, based on the genetic algorithm, water pollutant concentration prediction model and treatment cost prediction model, the principle underlying the optimal water pollution load distribution solution is as follows:

[0030] Based on the genetic algorithm, the pollutant emission allocation, discharge flow and pollutant emission concentration are encoded into chromosomes, and the initial population is randomly generated. Each individual in the population represents a set of pollutant emission allocation, discharge flow and pollutant emission concentration. Based on the water pollutant concentration prediction model and the treatment cost prediction model, the pollutant concentration and treatment cost are generated, and then the fitness function of each individual is generated;

[0031] Based on tournament selection, the individuals of the initial population are randomly arranged, and individuals are selected in pairs before and after the initial population after arrangement, and the fitness function values ​​of the individuals are calculated. The individual with the smaller fitness function value between the two is selected as the parent generation. The process is repeated until all individuals of the initial population are traversed, which is considered to be a round of selection. Crossover and mutation are randomly performed on the selected parent individuals to generate new offspring, and the offspring and parent generations are combined to form a new population. The fitness function value of each individual is calculated, and the individual with the lowest fitness function value is recorded as the current optimal solution. The above operation is regarded as one iteration, and the maximum number of iterations is set to 100. When the number of iterations reaches 100, or the fitness function value decreases by less than 5% in 10 consecutive iterations, the iteration is terminated, and the individual with the lowest current fitness function value is output. The corresponding pollutant emission allocation, discharge flow rate and pollutant emission concentration are the allocation plans of each sewage outlet in the future discharge time period.

[0032] The present invention also provides a multi-objective optimization water pollution load distribution system, which is used to implement the multi-objective optimization water pollution load distribution method mentioned above, and specifically includes:

[0033] The data acquisition module is used to divide the target water body into multiple pollution zones and obtain the pollutant discharge distribution, treatment cost, discharge flow rate and pollutant discharge concentration of each sewage outlet during the previous sewage discharge period, as well as the maximum water pollutant concentration of each pollution zone during the corresponding sewage discharge period;

[0034] The concentration prediction module is used to build a water pollutant concentration prediction model based on a deep neural network. The discharge flow and pollutant emission concentration of all sewage outlets in the same sewage discharge time period are used as input, and the maximum water pollutant concentration in the polluted area is used as a label to train the water pollutant concentration prediction model.

[0035] The emission calculation module is used to calculate the actual pollutant emissions, pollutant reduction and pollutant treatment load of each sewage outlet during the sewage discharge time period based on the pollutant emission allocation, sewage flow rate and pollutant concentration;

[0036] The cost prediction module is used to build a governance cost prediction model based on a deep neural network. The model takes the pollutant reduction, discharge flow rate, and pollutant emission concentration of each sewage outlet during the discharge period as input and the governance cost as a label to train the governance cost prediction model.

[0037] The comprehensive optimization module is used to obtain the total amount of pollutant discharge allocation in the future discharge period, and to generate a fitness function with the optimization objectives of minimizing the total cost of treatment, minimizing the concentration of water pollutants, and minimizing the pollutant treatment load. The maximum water pollutant concentration in each polluted area is less than the pollutant concentration standard, the sum of the pollutant discharge allocation of each sewage outlet is equal to the total amount of pollutant discharge allocation, and the pollutant treatment load of each sewage outlet is not greater than the maximum design load. Based on the genetic algorithm, the water pollutant concentration prediction model and the treatment cost prediction model, the optimal water pollution load distribution plan is obtained. The optimal water pollution load distribution plan is the pollutant discharge allocation, discharge flow and pollutant emission concentration of each sewage outlet in the future discharge period.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The present invention comprehensively considers the pollutant emission allocation, discharge flow and pollutant emission concentration; through regional division and comprehensive utilization of historical data, it can more accurately capture the spatiotemporal variation characteristics of water pollution, provide support for dynamic optimization, and through regional division and data collection, it can clarify the governance needs and characteristics of each pollution area, provide reliable input and constraint conditions for subsequent optimization algorithms, and help to achieve multi-objective systematic optimization; and by comprehensively considering the pollutant emission allocation, discharge flow and pollutant concentration, it can accurately calculate the actual pollutant emission, pollutant reduction and pollutant treatment load of each sewage outlet, thereby comprehensively reflecting the actual treatment capacity of the sewage outlet.

[0040] The present invention also trains a water pollutant concentration prediction model and a treatment cost prediction model through a deep learning network, which can accurately capture the complex dynamic relationship between discharge flow, pollutant concentration and target water pollutant concentration, and has higher prediction accuracy. The pollutant concentration and treatment cost data predicted by the model can directly guide the distribution of discharge load, provide an intuitive reference basis for decision makers, and avoid treatment errors or waste of resources due to uncertainty in traditional methods; by simultaneously optimizing the total treatment cost, water pollutant concentration and pollutant treatment load, taking into account environmental and economic benefits, and introducing genetic algorithms, it can efficiently converge to the optimal solution under multivariable and nonlinear conditions, make full use of historical data and real-time monitoring data, and optimize each sewage outlet. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a schematic diagram of a method flow in accordance with an embodiment of the present invention;

[0042] Figure 2 Schematic diagram of system modules according to an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0044] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0045] Example:

[0046] See also Figure 1 , the present invention provides a technical solution:

[0047] A multi-objective optimization method for water pollution load distribution includes the following steps:

[0048] Step 1: Divide the target water body into multiple pollution zones, and obtain the pollutant discharge distribution, treatment cost, discharge flow rate, and pollutant discharge concentration of each sewage outlet during the previous discharge period, as well as the maximum water pollutant concentration of each pollution zone during the corresponding discharge period;

[0049] In this embodiment, the pollutant emission allocation amount represents the allocation value of the total pollution emission to each sewage outlet, describes the actual pollutant emission situation of each sewage outlet in the previous sewage discharge time period, and is used to evaluate the emission capacity of the sewage outlet; the treatment cost represents the economic cost incurred by reducing the pollutant emission concentration of the sewage outlet during the sewage discharge time period, and is determined by collecting historical sewage treatment financial expenditure details; the sewage flow rate represents the sewage discharge volume containing pollutants from each sewage outlet per unit time, and an ultrasonic flow meter is installed at the sewage outlet for real-time monitoring; the pollutant emission concentration represents the concentration of the pollutant with the highest proportion in the wastewater at the sewage outlet during the sewage discharge time period, and an online pollutant concentration monitoring instrument is installed at the sewage outlet to obtain real-time data, and the online monitoring instrument includes a COD monitor and an ammonia nitrogen analyzer; the maximum water pollution concentration represents the maximum concentration of pollutants in the water body during the sewage discharge time period.

[0050] Step 2: Build a water pollutant concentration prediction model based on a deep neural network. The discharge flow and pollutant discharge concentration of all sewage outlets within the same discharge time period are used as input, and the maximum water pollutant concentration in the polluted area is used as a label to train the water pollutant concentration prediction model.

[0051] In this embodiment, the deep learning network structure for constructing the water pollutant concentration prediction model is:

[0052] Output layer: contains 2N neurons, which are used to input the discharge flow and pollutant emission concentration of each sewage outlet, where N represents the number of sewage outlets;

[0053] The first hidden layer: contains 64 neurons and is activated using the ReLU activation function;

[0054] The second hidden layer contains 32 neurons and is activated using the ReLU activation function.

[0055] The third hidden layer: contains 16 neurons and is activated using the ReLU activation function;

[0056] Output layer: contains N neurons, which are used to output the maximum concentration of water pollutants in each polluted area.

[0057] Historical sewage discharge flow and pollutant emission concentration data were collected, and 80% of the historical data were randomly selected as the training set to train the water pollutant concentration prediction model, and 20% were used as the validation set. The mean square error function was constructed to compare the error between the predicted water pollutant concentration and the actual water pollutant concentration to verify the prediction accuracy of the model. The smaller the mean square error function value, the closer the model prediction value is to the true value, and the better the model performance.

[0058] Step 3: Based on the pollutant emission allocation, discharge flow rate, and pollutant concentration, calculate the actual pollutant emissions, pollutant reduction, and pollutant treatment load for each discharge outlet during the discharge period;

[0059] In this embodiment, the principles for calculating the actual pollutant discharge, pollutant reduction, and pollutant treatment load of each sewage outlet during the sewage discharge period are as follows:

[0060] The formula for calculating actual pollutant emissions is:

[0061] M i =Q i ·C i ·T

[0062] Among them, M i represents the amount of pollutant discharged during the discharge period of the i-th sewage outlet, i represents the index of the sewage outlet, Q i represents the discharge flow of the i-th sewage outlet, C i represents the average pollutant concentration of the ith sewage outlet during the sewage discharge time period, and T represents the duration of the sewage discharge time period;

[0063] The sewage discharge flow rate represents the amount of sewage discharged per unit time, and the pollutant concentration represents the mass of pollutants contained in a unit volume of sewage. The product of the two represents the pollutant discharge per unit time, and multiplying it by the discharge time generates the actual pollutant discharge for that time period.

[0064] The formula for calculating pollutant reductions is:

[0065] D i =F i -M i

[0066] Among them, D i represents the pollutant reduction of the i-th sewage outlet, F i represents the pollutant emission allocation of the i-th sewage outlet;

[0067] The pollutant reduction amount refers to the pollutant emission reduction amount after sewage outlet treatment. The difference between the pollutant emission allocation amount and the actual emission amount is the pollutant emission reduction amount after sewage outlet treatment.

[0068] The formula for calculating pollutant treatment load is:

[0069]

[0070] Among them, L i Represents the treatment load of the i-th sewage outlet per unit time.

[0071] The pollutant treatment load represents the rate of pollutant reduction per unit time, and the pollutant reduction amount represents the total amount during the entire discharge time. Dividing it by the discharge time represents the treatment load.

[0072] Step 4: Build a governance cost prediction model based on a deep neural network, using the pollutant reduction, discharge flow rate, and pollutant emission concentration of each sewage outlet during the discharge period as input and the governance cost as a label to train the governance cost prediction model;

[0073] In this embodiment, the deep neural network structure for constructing the governance cost prediction model is:

[0074] Input layer: contains 3N neurons, which are used to input the pollutant reduction amount, discharge flow rate and pollutant emission concentration of each sewage outlet;

[0075] The first hidden layer: contains 64 neurons and is activated using the ReLU activation function;

[0076] The second hidden layer contains 32 neurons and is activated using the ReLU activation function.

[0077] The third hidden layer: contains 16 neurons and is activated using the ReLU activation function;

[0078] Output layer: contains N neurons, which are used to output the treatment cost of each sewage outlet during the sewage discharge period.

[0079] Historical data on pollutant reduction, discharge flow and pollutant emission concentration were collected, and 80% of the historical data were randomly selected as the training set to train the governance cost prediction model. The remaining 20% ​​of the data was used as the validation set. The mean square error function was constructed to compare the error between the predicted governance cost and the actual governance cost. The smaller the value of the mean square error function, the closer the predicted value is to the true value, and the better the model performance.

[0080] The cost of pollution control is not completely linear, but rather exhibits the characteristics of decreasing marginal costs. As the amount of pollutant reduction increases, the cost of pollution control will go through the following stages:

[0081] Initial stage: low-cost areas, where a large amount of pollutants can be reduced through simple measures such as basic wastewater treatment and reducing emission concentrations, and the cost of treatment increases linearly with the amount of reduction;

[0082] Transition stage: Increasing cost zone, further reduction of pollutants requires more complex technologies, the unit reduction cost increases, and it shows a nonlinear growth trend;

[0083] Deep governance stage: In high-cost areas, when reduction approaches its limit, costs rise sharply and may even exceed economic feasibility;

[0084] Step 5: Obtain the total amount of pollutant discharge allocation in the future pollution discharge period, and generate a fitness function with the optimization objectives of minimizing the total cost of treatment, minimizing the water pollutant concentration, and minimizing the pollutant treatment load. With the maximum water pollutant concentration in each pollution area less than the pollutant concentration standard, the sum of the pollutant discharge allocation of each sewage outlet equal to the total amount of pollutant discharge allocation, and the pollutant treatment load of each sewage outlet not greater than the maximum design load as constraints, based on the genetic algorithm, the water pollutant concentration prediction model and the treatment cost prediction model, the optimal water pollution load allocation scheme is obtained. The optimal water pollution load allocation scheme is the pollutant discharge allocation, discharge flow and pollutant discharge concentration of each sewage outlet in the future pollution discharge period.

[0085] In this embodiment, the optimization objectives are to minimize the total cost of treatment, minimize the concentration of water pollutants, and minimize the pollutant treatment load. The generated fitness function is:

[0086]

[0087] Where X represents the total cost of treatment, i represents the index of the sewage outlet, and i∈[1,N], N represents the number of sewage outlets, Si represents the treatment cost of the ith sewage outlet output by the treatment cost prediction model, X max represents the upper limit of the total cost of governance, j represents the index of the polluted area, and j∈[1,H], H represents the number of polluted areas divided into water bodies, P j It represents the maximum pollutant concentration of the jth polluted area output by the water pollutant concentration prediction model, P j,std represents the expected value of pollutant concentration in the jth polluted area, L i,max represents the maximum design load of the i-th sewage outlet, w1, w2, and w3 represent weight coefficients, and w1+w2+w3=1, w2>w3>w1.

[0088] represents the normalized treatment of the total governance cost, X max It indicates the total cost of treatment when all sewage outlets have to bear the maximum design load, max(0,P j -P j,std ) represents the excess of the water pollutant concentration in each polluted area relative to the expected value of the pollutant concentration. If it does not exceed the standard, the value is 0. If it exceeds the standard, the excess values ​​of all areas are normalized and accumulated to obtain the optimization target of the overall concentration of the polluted area; max(0,L i -L i,max ) represents the excess of the treatment load of each sewage outlet relative to the maximum design load. If the load does not exceed the limit, the value is 0. If the load exceeds the limit, all excess amounts are normalized and accumulated as the optimization target of the sewage outlet treatment load.

[0089] In water pollution allocation and management, the most important goal is to reduce the concentration of pollutants. It is necessary to prioritize ensuring that the concentration of pollutants in the polluted area is lower than the expected value, so w2 takes the highest value; the pollutant treatment load directly determines the working pressure of the sewage outlet. If the load exceeds the standard, it may lead to a decrease in sewage discharge efficiency, thereby affecting the actual treatment effect, so w3 takes the second value; the total cost of management is a relatively flexible goal, which can be adjusted through different management technologies, optimization plans and funding allocation strategies. It has a certain degree of flexibility and can be gradually optimized through long-term planning. Therefore, w1 takes the smallest value, and the specific weights are: w1=0.2, w2=0.5, w3=0.3.

[0090] Based on the genetic algorithm, water pollutant concentration prediction model and treatment cost prediction model, the principle of obtaining the optimal water pollution load distribution plan is as follows:

[0091] Based on the genetic algorithm, the pollutant emission allocation, discharge flow and pollutant emission concentration are encoded into chromosomes, and the initial population is randomly generated. Each individual in the population represents a set of pollutant emission allocation, discharge flow and pollutant emission concentration. Based on the water pollutant concentration prediction model and the treatment cost prediction model, the pollutant concentration and treatment cost are generated, and then the fitness function of each individual is generated;

[0092] Based on tournament selection, the individuals of the initial population are randomly arranged, and individuals are selected in pairs before and after the initial population after arrangement, and the fitness function values ​​of the individuals are calculated. The individual with the smaller fitness function value between the two is selected as the parent generation. The process is repeated until all individuals of the initial population are traversed, which is considered to be a round of selection. Crossover and mutation are randomly performed on the selected parent individuals to generate new offspring, and the offspring and parent generations are combined to form a new population. The fitness function value of each individual is calculated, and the individual with the lowest fitness function value is recorded as the current optimal solution. The above operation is regarded as one iteration, and the maximum number of iterations is set to 100. When the number of iterations reaches 100, or the fitness function value decreases by less than 5% in 10 consecutive iterations, the iteration is terminated, and the individual with the lowest current fitness function value is output. The corresponding pollutant emission allocation, discharge flow rate and pollutant emission concentration are the allocation plans of each sewage outlet in the future discharge time period.

[0093] See also Figure 2 The present invention also provides a multi-objective optimization water pollution load distribution system, which is used to implement the multi-objective optimization water pollution load distribution method, specifically including:

[0094] The data acquisition module is used to divide the target water body into multiple pollution zones and obtain the pollutant discharge distribution, treatment cost, discharge flow rate and pollutant discharge concentration of each sewage outlet during the previous sewage discharge period, as well as the maximum water pollutant concentration of each pollution zone during the corresponding sewage discharge period;

[0095] The concentration prediction module is used to build a water pollutant concentration prediction model based on a deep neural network. The discharge flow and pollutant emission concentration of all sewage outlets in the same sewage discharge time period are used as input, and the maximum water pollutant concentration in the polluted area is used as a label to train the water pollutant concentration prediction model.

[0096] The emission calculation module is used to calculate the actual pollutant emissions, pollutant reduction and pollutant treatment load of each sewage outlet during the sewage discharge time period based on the pollutant emission allocation, sewage flow rate and pollutant concentration;

[0097] The cost prediction module is used to build a governance cost prediction model based on a deep neural network. The model takes the pollutant reduction, discharge flow rate, and pollutant emission concentration of each sewage outlet during the discharge period as input and the governance cost as a label to train the governance cost prediction model.

[0098] The comprehensive optimization module is used to obtain the total amount of pollutant discharge allocation in the future discharge period, and to generate a fitness function with the optimization objectives of minimizing the total cost of treatment, minimizing the concentration of water pollutants, and minimizing the pollutant treatment load. The maximum water pollutant concentration in each polluted area is less than the pollutant concentration standard, the sum of the pollutant discharge allocation of each sewage outlet is equal to the total amount of pollutant discharge allocation, and the pollutant treatment load of each sewage outlet is not greater than the maximum design load. Based on the genetic algorithm, the water pollutant concentration prediction model and the treatment cost prediction model, the optimal water pollution load distribution plan is obtained. The optimal water pollution load distribution plan is the pollutant discharge allocation, discharge flow and pollutant emission concentration of each sewage outlet in the future discharge period.

[0099] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0100] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0101] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0102] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A multi-objective optimization method for water pollution load distribution, characterized in that: The specific steps include: Step 1: Divide the target water body into multiple pollution zones, and obtain the pollutant discharge distribution, treatment cost, discharge flow rate, and pollutant discharge concentration of each sewage outlet during the previous discharge period, as well as the maximum water pollutant concentration of each pollution zone during the corresponding discharge period; Step 2: Build a water pollutant concentration prediction model based on a deep neural network. The discharge flow and pollutant discharge concentration of all sewage outlets within the same discharge time period are used as input, and the maximum water pollutant concentration in the polluted area is used as a label to train the water pollutant concentration prediction model. Step 3: Based on the pollutant emission allocation, discharge flow rate, and pollutant concentration, calculate the actual pollutant emissions, pollutant reduction, and pollutant treatment load for each discharge outlet during the discharge period; Step 4: Build a governance cost prediction model based on a deep neural network, using the pollutant reduction, discharge flow rate, and pollutant emission concentration of each sewage outlet during the discharge period as input and the governance cost as a label to train the governance cost prediction model; Step 5: Obtain the total amount of pollutant emissions allocated during the future pollution discharge period, and generate a fitness function with the optimization objectives of minimizing the total cost of treatment, minimizing the water pollutant concentration, and minimizing the pollutant treatment load. With the constraints that the maximum water pollutant concentration in each polluted area is less than the pollutant concentration standard, the sum of the pollutant emission allocations of each sewage outlet is equal to the total amount of pollutant emission allocations, and the pollutant treatment load of each sewage outlet is no more than the maximum design load, the optimal water pollution load allocation scheme is obtained based on the genetic algorithm, the water pollutant concentration prediction model, and the treatment cost prediction model. The optimal water pollution load allocation scheme is the pollutant emission allocation, discharge flow, and pollutant emission concentration of each sewage outlet during the future pollution discharge period; Taking minimization of total treatment cost, minimization of water pollutant concentration, and minimization of pollutant treatment load as optimization objectives, the generated fitness function is: Where X represents the total cost of treatment, i represents the index of the sewage outlet, and i∈[1,N], N represents the number of sewage outlets, S i represents the treatment cost of the ith sewage outlet output by the treatment cost prediction model, X max represents the upper limit of the total cost of governance, j represents the index of the polluted area, and j∈[1,H], H represents the number of polluted areas divided into water bodies, P j It represents the maximum pollutant concentration of the jth polluted area output by the water pollutant concentration prediction model, P j,std represents the expected value of pollutant concentration in the jth polluted area, L i,max represents the maximum design load of the i-th sewage outlet, w1, w2, and w3 represent weight coefficients, and w1+w2+w3=1, w2>w3>w1.

2. The multi-objective optimization method for water pollution load distribution according to claim 1, characterized in that: The pollutant emission allocation in step 1 is the pollutant value allocated to each sewage outlet; the treatment cost represents the economic cost incurred by reducing the pollutant emission concentration at the sewage outlet during the sewage discharge time period, which is determined by collecting historical sewage treatment financial expenditure details; the sewage flow rate represents the amount of sewage containing pollutants discharged from each sewage outlet per unit time, and an ultrasonic flow meter is installed at the sewage outlet for real-time monitoring; the pollutant emission concentration represents the concentration of the pollutant with the highest proportion in the wastewater at the sewage outlet during the sewage discharge time period, and an online pollutant concentration monitoring instrument is installed at the sewage outlet to obtain real-time data, and the online monitoring instrument includes a COD monitor and an ammonia nitrogen analyzer; the maximum water pollution concentration represents the maximum concentration of pollutants in the water body during the sewage discharge time period.

3. The multi-objective optimization method for water pollution load distribution according to claim 1, characterized in that: The principle for calculating the actual pollutant discharge, pollutant reduction and pollutant treatment load of each sewage outlet during the sewage discharge period in step 3 is as follows: The formula for calculating actual pollutant emissions is: M i =Q i ·C i . Among them, M i represents the pollutant emissions from the i-th sewage outlet during the discharge period, i represents the index of the sewage outlet, Q i represents the discharge flow of the i-th sewage outlet, C i represents the average pollutant concentration of the ith sewage outlet during the sewage discharge time period, and T represents the duration of the sewage discharge time period; The formula for calculating pollutant reductions is: D i =F i -M i Among them, D i represents the pollutant reduction of the i-th sewage outlet, F i represents the pollutant emission allocation of the i-th sewage outlet; The formula for calculating pollutant treatment load is: Among them, L i Represents the treatment load of the i-th sewage outlet per unit time.

4. The multi-objective optimization method for water pollution load distribution according to claim 1, characterized in that: The principle underlying the optimal water pollution load distribution solution obtained in step 5 based on the genetic algorithm, the water pollutant concentration prediction model, and the treatment cost prediction model is: Based on the genetic algorithm, the pollutant emission allocation, discharge flow and pollutant emission concentration are encoded into chromosomes, and the initial population is randomly generated. Each individual in the population represents a set of pollutant emission allocation, discharge flow and pollutant emission concentration. Based on the water pollutant concentration prediction model and the treatment cost prediction model, the pollutant concentration and treatment cost are generated, and then the fitness function of each individual is generated; Based on tournament selection, the individuals of the initial population are randomly arranged, and individuals are selected in pairs before and after the initial population after arrangement, and the fitness function values ​​of the individuals are calculated. The individual with the smaller fitness function value between the two is selected as the parent generation. The process is repeated until all individuals of the initial population are traversed, which is considered to be a round of selection. Crossover and mutation are randomly performed on the selected parent individuals to generate new offspring, and the offspring and parent generations are combined to form a new population. The fitness function value of each individual is calculated, and the individual with the lowest fitness function value is recorded as the current optimal solution. The above operation is regarded as one iteration, and the maximum number of iterations is set to 100. When the number of iterations reaches 100, or the fitness function value decreases by less than 5% in 10 consecutive iterations, the iteration is terminated, and the individual with the lowest current fitness function value is output. The corresponding pollutant emission allocation, discharge flow rate and pollutant emission concentration are the allocation plans of each sewage outlet in the future discharge time period.

5. A multi-objective optimization water pollution load distribution system, characterized by: The system is used to implement the multi-objective optimization water pollution load distribution method according to any one of claims 1 to 4, specifically comprising: The data acquisition module is used to divide the target water body into multiple pollution zones and obtain the pollutant discharge distribution, treatment cost, discharge flow rate and pollutant discharge concentration of each sewage outlet during the previous sewage discharge period, as well as the maximum water pollutant concentration of each pollution zone during the corresponding sewage discharge period; The concentration prediction module is used to build a water pollutant concentration prediction model based on a deep neural network. The discharge flow and pollutant emission concentration of all sewage outlets in the same sewage discharge time period are used as input, and the maximum water pollutant concentration in the polluted area is used as a label to train the water pollutant concentration prediction model. The emission calculation module is used to calculate the actual pollutant emissions, pollutant reduction and pollutant treatment load of each sewage outlet during the sewage discharge time period based on the pollutant emission allocation, sewage flow rate and pollutant concentration; The cost prediction module is used to build a governance cost prediction model based on a deep neural network. The model takes the pollutant reduction, discharge flow rate, and pollutant emission concentration of each sewage outlet during the discharge period as input and the governance cost as a label to train the governance cost prediction model. The comprehensive optimization module is used to obtain the total amount of pollutant discharge allocation in the future discharge period, and to generate a fitness function with the optimization objectives of minimizing the total cost of treatment, minimizing the concentration of water pollutants, and minimizing the pollutant treatment load. The maximum water pollutant concentration in each polluted area is less than the pollutant concentration standard, the sum of the pollutant discharge allocation of each sewage outlet is equal to the total amount of pollutant discharge allocation, and the pollutant treatment load of each sewage outlet is not greater than the maximum design load. Based on the genetic algorithm, the water pollutant concentration prediction model and the treatment cost prediction model, the optimal water pollution load distribution plan is obtained. The optimal water pollution load distribution plan is the pollutant discharge allocation, discharge flow and pollutant emission concentration of each sewage outlet in the future discharge period.

Images