Neural network-based dynamic multi-objective optimization control method for wastewater treatment process

Abstract

The application provides a sewage treatment process dynamic multi-objective optimization control method based on a neural network, comprising: obtaining characteristic variables of effluent quality and energy consumption of a sewage treatment process; constructing an optimized effluent quality model and an optimized energy consumption model according to the characteristic variables based on an echo state network; performing optimization on the optimized effluent quality model and the optimized energy consumption model by a dynamic multi-objective optimization algorithm to obtain a target effluent quality model and a target energy consumption model; determining optimal dissolved oxygen concentration and optimal nitrate nitrogen concentration in the target effluent quality model and the target energy consumption model; and performing dynamic tracking control on the sewage treatment process by using a single neuron adaptive controller according to the optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration based on a control formula. The application solves the problems of difficulty in dynamic optimization control of the sewage treatment process and increase of energy consumption and operation cost of the sewage treatment process in the prior art.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, and in particular to a dynamic multi-objective optimization control method for wastewater treatment processes based on neural networks. Background Technology

[0002] Urban wastewater treatment involves complex biochemical reactions and is characterized by dynamism, nonlinearity, and uncertainty. The core objective of wastewater treatment is to remove or reduce harmful substances in wastewater to safe levels, ensuring that effluent quality meets environmental standards. With increasing wastewater discharge volumes and increasingly stringent emission standards, wastewater treatment costs are constantly rising. Therefore, ensuring effluent quality meets standards while reducing operational energy consumption is a significant task and challenge for wastewater treatment processes.

[0003] To adapt to dynamic wastewater treatment processes, improve treatment efficiency, and reduce operating costs, dynamic optimization control methods for wastewater treatment processes have emerged. To verify the effectiveness of different optimization control methods in wastewater treatment processes, the International Water Quality Association and the European Union Organization for Scientific and Technical Cooperation jointly developed Benchmark Simulation Model No. 1 (BSM1). BSM1 consists of a biological reactor and a secondary sedimentation tank. The biological reactor comprises five units: Zone 1, Zone 2, Zone 3, Zone 4, and Zone 5. The first two units are anoxic zones, primarily where denitrification occurs, while the latter three units are aerobic zones, primarily where nitrification occurs. Wastewater is clarified in the secondary sedimentation tank; the supernatant flows out, and part of the lower sludge flows back into the anoxic zone via external recirculation, while the remainder is discharged as excess sludge. In the optimization control process, an accurate description of the dynamic mapping relationship between effluent quality, operating energy consumption, and their key variables is the foundation and prerequisite for achieving optimized control of the wastewater treatment process. In existing technologies, dynamic optimization control of wastewater treatment processes is difficult to achieve. Furthermore, in actual wastewater treatment processes, the changes in dissolved oxygen concentration in the fifth zone and nitrate nitrogen concentration in the second zone directly affect the nitrification and denitrification processes, thereby impacting the energy consumption and operating costs of the wastewater treatment process, leading to an increase in energy consumption and operating costs.

[0004] This invention conducts dynamic optimization control experiments for wastewater treatment based on the BSM1 model. It dynamically obtains the optimized setpoints for the dissolved oxygen concentration in the fifth zone and the nitrate nitrogen concentration in the second zone, and utilizes a single-neuron adaptive controller to achieve effective tracking control of the optimized setpoints. The dissolved oxygen concentration in the fifth zone of the aerobic zone is represented by S. O The nitrate nitrogen concentration in the second anaerobic zone is expressed as S. NO Oxygen transfer coefficient K in the fifth zone L a5 is S OThe control parameters, and the internal return flow Q from the fifth partition to the second partition. a It is S NO The control parameters. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a dynamic multi-objective optimization control method for wastewater treatment processes based on neural networks. This invention solves the problems of difficulty in achieving dynamic optimization control of wastewater treatment processes and the increase in energy consumption and operating costs in existing technologies.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A dynamic multi-objective optimization control method for wastewater treatment processes based on neural networks, comprising:

[0008] The characteristic variables of effluent quality and energy consumption in the wastewater treatment process are obtained; wherein, the characteristic variables include: dissolved oxygen concentration, nitrate nitrogen concentration, suspended solids concentration, ammonia nitrogen concentration and influent flow rate;

[0009] Based on echo state networks, an effluent quality model and an energy consumption model to be optimized are constructed according to characteristic variables.

[0010] Based on the dynamic multi-objective optimization algorithm of two-space prediction, the effluent water quality model and the energy consumption model to be optimized are optimized to obtain the target effluent water quality model and the target energy consumption model.

[0011] Determine the optimal dissolved oxygen concentration and optimal nitrate nitrogen concentration in the target effluent water quality model and the target energy consumption model;

[0012] Based on the control formula, a single-neuron adaptive controller is used to dynamically track and control the wastewater treatment process according to the optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration.

[0013] Preferably, the energy consumption model is as follows:

[0014]

[0015] f1(x(t)) is the energy consumption model at time t, with the input variable being xt = S. O t S NO t,MLSS t S NH t,Q in (t], where S O (t) represents the dissolved oxygen concentration at time t, S NO (t) represents the nitrate nitrogen concentration at time t, MLSS(t) represents the suspended solids concentration at time t, and S NH (t) represents the ammonia nitrogen concentration at time t, Q in(t) represents the inflow rate at time t, and h(·) represents the sigmoid activation function. and W1 is the input weight matrix, and W2 is the internal weight matrix of the reserve pool. and H1 and H2 are the output weight matrix, H1 and H2 are the internal state matrices, s1(t) and s2(t) are the echo state matrices at time t, and Y1 and Y2 are the target output matrices.

[0016] Preferably, the effluent water quality model is:

[0017]

[0018] Where f2(x(t)) is the effluent water quality model at time t.

[0019] Preferably, a dynamic multi-objective optimization algorithm based on two-space prediction is used to optimize the effluent water quality model and the energy consumption model to be optimized, resulting in a target effluent water quality model and a target energy consumption model, including:

[0020] Set the population size to N∈[50,150] and the maximum number of iterations to G. max ∈[40,50];

[0021] Using the energy consumption model and the effluent water quality model to be optimized as optimization objectives, and the dissolved oxygen concentration in the fifth zone and the nitrate nitrogen concentration in the second zone of the BSM1 model as decision variables, the population is initialized based on a dynamic multi-objective optimization algorithm with two spatial predictions, and the objective function value is calculated to obtain the target energy consumption model and the target effluent water quality model.

[0022] Preferably, a dynamic multi-objective optimization algorithm based on two-space prediction is used to initialize the population, calculate the objective function value, and obtain the target energy consumption model and the target effluent water quality model, including:

[0023] Obtain the environment vector;

[0024] The environment vector is normalized to obtain a normalized vector;

[0025] If the Euclidean distance between the normalized vectors is greater than the environmental change threshold, then the optimized setpoints for dissolved oxygen and nitrate nitrogen concentrations are calculated using a dynamic multi-objective optimization algorithm based on two-space prediction to obtain the target energy consumption model and the target effluent water quality model. If not, then the initial population at the next moment is predicted based on the initial population at the current moment and the initial population at the previous moment to obtain the initial population groups at each moment. The initial population groups at each moment are then optimized based on the decomposition multi-objective evolutionary algorithm until the multi-objective evolutionary algorithm reaches the maximum number of iterations to obtain the target energy consumption model and the target effluent water quality model.

[0026] Preferably, the initial population at the next time step is predicted based on the initial population at the current time step and the initial population at the previous time step, thus obtaining the initial population at each time step, including:

[0027] Within the decision space, based on the initial population at the current moment and the initial population at the previous moment, half of the initial population group is obtained;

[0028] In the target space, based on the initial population at the current moment and the initial population at the previous moment, obtain the other half of the initial population.

[0029] The first half of the initial seed cluster and the second half of the initial seed cluster are merged to obtain the initial seed clusters at each time point.

[0030] Preferably, determining the optimal dissolved oxygen concentration and optimal nitrate nitrogen concentration in the target effluent water quality model and the target energy consumption model includes:

[0031] Obtain the solution sets of dissolved oxygen concentration and optimal nitrate nitrogen concentration from the target effluent water quality model and the target energy consumption model;

[0032] Based on the inflection point calculation formula, the inflection points in the solution sets of dissolved oxygen concentration and optimal nitrate nitrogen concentration are calculated;

[0033] The optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration are determined based on the inflection point.

[0034] Preferably, the inflection point calculation formula is:

[0035]

[0036] The solution with the largest Dis value is the inflection point, where a, b, and c are coefficients determined by two boundary points in the non-dominated solution set.

[0037] Preferably, the single-neuron adaptive controller is:

[0038]

[0039] K1 is the neuron scaling factor, satisfying K1>0, w i (t) represents the weighting coefficient.

[0040] Preferably, the control formula is:

[0041]

[0042] Where, γ i >0 represents the learning rate of the neuron, v1(t), v2(t), and v3(t) represent the inputs of the neuron, and Δu(t) = [ΔK] L a5(t),ΔQa (t)] T ΔK L a5(t) is the change in the oxygen transfer coefficient of the fifth region at time t, ΔQ a e(t) is the change in return flow rate at time t, and e(t) = z * (t)-z(t) represents the control error between the actual output and the optimized setpoint, z * (t)=[S O * (t),S NO * [(t)] represents the optimal setpoints for dissolved oxygen concentration and nitrate nitrogen concentration at time t, z(t) = [S O (t),S NO [(t)] represents the actual output dissolved oxygen concentration and nitrate nitrogen concentration at time t.

[0043] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0044] This invention provides a dynamic multi-objective optimization control method for wastewater treatment processes based on neural networks, comprising: acquiring characteristic variables of effluent water quality and energy consumption in the wastewater treatment process; wherein, the characteristic variables include: dissolved oxygen concentration, nitrate nitrogen concentration, suspended solids concentration, ammonia nitrogen concentration, and influent flow rate; constructing an effluent water quality model and an energy consumption model to be optimized based on echo state networks and the characteristic variables; optimizing the effluent water quality model and the energy consumption model to be optimized based on a two-space prediction dynamic multi-objective optimization algorithm to obtain a target effluent water quality model and a target energy consumption model; determining the optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration in the target effluent water quality model and the target energy consumption model; and using a single-neuron adaptive controller to dynamically track and control the wastewater treatment process based on control formulas and the optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration. This invention can not only calculate the optimal setpoints of dissolved oxygen concentration and nitrate nitrogen concentration in real time, but also achieve accurate tracking of the optimal setpoints of dissolved oxygen concentration and nitrate nitrogen concentration. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 A flowchart of a dynamic multi-objective optimization control method for wastewater treatment based on neural networks is provided in this embodiment of the invention.

[0047] Figure 2 This is an overall structural diagram of the dynamic optimization control system provided in an embodiment of the present invention;

[0048] Figure 3 The dissolved oxygen concentration setpoint and tracking result graph provided in the embodiments of the present invention;

[0049] Figure 4 This is a diagram illustrating the dissolved oxygen concentration tracking and control error provided in an embodiment of the present invention.

[0050] Figure 5 The following is a graph showing the setpoint and tracking results of nitrate nitrogen concentration provided in an embodiment of the present invention;

[0051] Figure 6 Error diagram for nitrate nitrogen concentration tracking and control provided in the embodiments of the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] The purpose of this invention is to provide a dynamic multi-objective optimization control method for wastewater treatment processes based on neural networks. This invention solves the problems of difficulty in achieving dynamic optimization control of wastewater treatment processes and the increase in energy consumption and operating costs in existing technologies.

[0054] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] like Figure 1 As shown, this invention provides a dynamic multi-objective optimization control method for wastewater treatment processes based on neural networks, comprising:

[0056] Step 100: Obtain characteristic variables of effluent quality and energy consumption in the wastewater treatment process; wherein, the characteristic variables include: dissolved oxygen concentration, nitrate nitrogen concentration, suspended solids concentration, ammonia nitrogen concentration, and influent flow rate;

[0057] Specifically, the dynamic characteristics and operational data of the wastewater treatment process are analyzed to obtain key characteristic variables affecting effluent quality and energy consumption: dissolved oxygen concentration S. O Nitrate nitrogen concentration S NO MLSS (Mixed Solids Suspended) and S (Ammonia Nitrogen Concentration) NH and inflow rate Q in ;

[0058] Step 200: Based on the echo state network, construct the effluent water quality model and the energy consumption model to be optimized according to the feature variables;

[0059] Step 300: Based on the dynamic multi-objective optimization algorithm of two-space prediction, optimize the effluent water quality model and the energy consumption model to be optimized to obtain the target effluent water quality model and the target energy consumption model.

[0060] Step 400: Determine the optimal dissolved oxygen concentration and optimal nitrate nitrogen concentration in the target effluent water quality model and the target energy consumption model;

[0061] Step 500: Based on the control formula, and according to the optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration, a single neuron adaptive controller is used to dynamically track and control the wastewater treatment process.

[0062] Furthermore, such as Figure 2 As shown, the energy consumption model is as follows:

[0063]

[0064] f1(x(t)) is the energy consumption model at time t, with the input variable being xt = S. O t S NO t,MLSS t S NH t,Q in (t], where S O (t) represents the dissolved oxygen concentration at time t, S NO (t) represents the nitrate nitrogen concentration at time t, MLSS(t) represents the suspended solids concentration at time t, and S NH (t) represents the ammonia nitrogen concentration at time t, Q in (t) represents the inflow rate at time t, and h(·) represents the sigmoid activation function. and W1 is the input weight matrix, and W2 is the weight matrix within the reserve pool. Both are randomly generated before training and remain unchanged once generated. and H1 and H2 are the output weight matrix, H1 and H2 are the internal state matrices, s1(t) and s2(t) are the echo state matrices at time t, and Y1 and Y2 are the target output matrices.

[0065] Furthermore, the effluent water quality model is as follows:

[0066]

[0067] Where f2(x(t)) is the effluent water quality model at time t.

[0068] Furthermore, based on a dynamic multi-objective optimization algorithm using two-space prediction, the effluent water quality model and the energy consumption model to be optimized are optimized to obtain the target effluent water quality model and the target energy consumption model, including:

[0069] Set the population size to N∈[50,150] and the maximum number of iterations to G. max ∈[40,50];

[0070] Using the energy consumption model and the effluent water quality model to be optimized as optimization objectives, and the dissolved oxygen concentration in the fifth zone and the nitrate nitrogen concentration in the second zone as decision variables, the population is initialized based on a dynamic multi-objective optimization algorithm with two spatial predictions, and the objective function value is calculated to obtain the target energy consumption model and the target effluent water quality model.

[0071] Specifically, the dynamic multi-objective optimization algorithm based on two-space prediction is equivalent to the two-space prediction strategy shown in the attached figure.

[0072] Specifically, the population size is set to N = 100, and the maximum number of iterations is Gmax = 40;

[0073] The energy consumption model and the effluent water quality model are set as optimization objectives, and the decision variables are the dissolved oxygen concentration in the fifth zone and the nitrate nitrogen concentration in the second zone, respectively. The population is initialized and the objective function value is calculated.

[0074] Furthermore, based on a dynamic multi-objective optimization algorithm using two-space prediction, the population is initialized, the objective function value is calculated, and the target energy consumption model and the target effluent water quality model are obtained, including:

[0075] Obtain the environment vector;

[0076] The environment vector is normalized to obtain a normalized vector;

[0077] If the Euclidean distance between the normalized vectors is greater than the environmental change threshold, then the optimized setpoints for dissolved oxygen and nitrate nitrogen concentrations are calculated using a dynamic multi-objective optimization algorithm based on two-space prediction to obtain the target energy consumption model and the target effluent water quality model. If not, then the initial population at the next moment is predicted based on the initial population at the current moment and the initial population at the previous moment to obtain the initial population groups at each moment. The initial population groups at each moment are then optimized based on the decomposition multi-objective evolutionary algorithm until the multi-objective evolutionary algorithm reaches the maximum number of iterations to obtain the target energy consumption model and the target effluent water quality model.

[0078] Specifically, the environment vector is set as [MLSS(t), S NH [(t)], where MLSS(t) is the current concentration of suspended solids in the effluent, S NH(t) represents the current ammonia nitrogen concentration in the effluent; the optimization cycle is set to T=2, meaning environmental monitoring is performed every 2 hours; environmental vectors need to be normalized, and when the Euclidean distance between environmental vectors is greater than the set environmental change threshold... hour, Setting it to 0.1 indicates that the environment has changed; when the environment changes, the dynamic multi-objective optimization algorithm based on two-space prediction is used to obtain the optimized setpoints for dissolved oxygen and nitrate nitrogen concentrations; when the environment does not change, the initial population is directly optimized using the decomposition-based multi-objective evolutionary algorithm MOEA / D.

[0079] The initial population is optimized using the decomposition-based multi-objective evolutionary algorithm MOEA / D to obtain a set of Pareto optimal solutions. This patent selects the weighted sum method as the decomposition method for MOEA / D.

[0080] The weighted summation method mainly assigns a weight value λ to each subproblem. i =(λ i1 ,λ i2 ), i∈1,...,N, and λ i1 +λ i2 =1, using the weight values, the multi-objective optimization problem can be transformed into the following problem:

[0081]

[0082] By changing λ, a set of Pareto optimal solutions can be obtained.

[0083] Determine if the algorithm has reached the set maximum number of evolution iterations G. max If the target is reached, the iterative evolution process terminates; if not, optimization continues.

[0084] Furthermore, based on the initial population at the current moment and the initial population at the previous moment, the initial population at the next moment is predicted, resulting in the initial population at each moment, including:

[0085] Within the decision space, based on the initial population at the current moment and the initial population at the previous moment, half of the initial population group is obtained;

[0086] In the target space, based on the initial population at the current moment and the initial population at the previous moment, obtain the other half of the initial population.

[0087] The first half of the initial seed cluster and the second half of the initial seed cluster are merged to obtain the initial seed clusters at each time point.

[0088] Specifically, the initial population of the new environment consists of two parts: first, in the decision space, 50% of the initial population is generated by establishing a denoising autoencoder prediction model; second, in the target space, the desired target vector is mapped back to the decision space by establishing an inverse model to generate the remaining 50% of the initial population.

[0089] Establish a denoising autoencoder model: When the environment changes, the Pareto optimal solutions obtained at environment t and t-1 are used to predict the initial population at time t+1. The Pareto optimal solutions for environments t and t-1 are represented as POS, respectively. t and POS t-1 Randomly select POS t and POS t-1 Half of the Pareto optimal solutions can be expressed as follows: and Where N t This indicates the number of solutions. Then, a denoising autoencoder is used for prediction, and settings are configured... As the input P of the denoising autoencoder, As the output Q, M is from arrive The learning mapping, M, is calculated using the following formula:

[0090] M = (QP) T (PP) T ) -1 ;

[0091] Where T is the matrix transpose operation, the Pareto optimal solution at time t+1 can be expressed as:

[0092]

[0093] Finally, the Pareto optimal solution based on the autoencoder model is obtained, denoted as Pop. dec .

[0094] Establish an inverse model for demapping: POS t and POS t-1 The remaining half of the solutions are represented as follows: and Then, calculate and The objective function values ​​are respectively expressed as: and use and Directly predict individuals in the target space. First, calculate... and center point F t and F t-1Then, predictions are made for individuals in the target space:

[0095]

[0096] Where δ is the variance perturbation, which is used to expand the search area and make the new population closer to the true point.

[0097] Given The predicted value is a set of target vectors, therefore it is necessary to... The solution is mapped back to the decision space to obtain a new Pareto optimal solution. Since radial basis function neural networks are easy to train and have excellent nonlinear fitting capabilities, they can map any complex nonlinear relationship. Therefore, this invention uses radial basis function neural networks to construct a nonlinear inverse model.

[0098] The implementation steps are as follows: and The inverse model is trained using the input and output as respectively, where, i∈1,...,N t Therefore, two inverse models need to be established. The obtained... and Divided into N t For each training data pair, the training samples for the first inverse model are... For the second inverse model, the training samples are Choosing the Gaussian function as the activation function for the radial basis function neural network, it is expressed as:

[0099]

[0100] Where σ > 0 is the variance of the activation function, and C is the center of the Gaussian function. Therefore, the output of the radial basis function neural network is:

[0101]

[0102] Where w j It is the weight of the j-th neuron.

[0103] Finally, the trained inverse model is used to obtain... The mapped solution in the decision space is considered as a predicted solution based on the target space, denoted as Pop. obj .

[0104] Pop obtained from the decision space dec The solution Pop obtained from the target space obj The population is merged to form the initial population at time t+1.

[0105] Furthermore, determining the optimal dissolved oxygen concentration and optimal nitrate nitrogen concentration in the target effluent water quality model and the target energy consumption model includes:

[0106] Obtain the solution sets of dissolved oxygen concentration and optimal nitrate nitrogen concentration from the target effluent water quality model and the target energy consumption model;

[0107] Based on the inflection point calculation formula, the inflection points in the solution sets of dissolved oxygen concentration and optimal nitrate nitrogen concentration are calculated;

[0108] The optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration are determined based on the inflection point.

[0109] Specifically, the optimal setpoint for the controller is selected as follows:

[0110] The optimal solutions on the Pareto front are not mutually exclusive, and the specific choice of the optimal solution directly affects the optimized control performance of the wastewater treatment process. Taking all factors into consideration, the inflection point among the Pareto optimal solutions is selected as the final setpoint for the controller. This invention employs a simple method to find inflection points, which determines whether a solution is an inflection point by calculating the distance from each solution to the extremum line. For a bi-objective optimization problem, the extremum line L is defined as follows:

[0111] L: a.f1+b.f2+c=0;

[0112] Where a, b, and c can be determined by two extreme points in the non-dominated solution set. Therefore, for a solution x with coordinates (f1(x), f2(x)), its distance to the extreme line L can be calculated as follows:

[0113]

[0114] The inflection point represents the solution with the largest Dis value in the non-dominated solution set. That is, the solution corresponding to the largest Dis value is selected as the optimal setting value, which can be represented as z. * (t)=[S O * (t),S NO * (t)].

[0115] Furthermore, the tracking and control of dissolved oxygen concentration and nitrate nitrogen concentration:

[0116] After obtaining the optimized setpoints for dissolved oxygen concentration and nitrate nitrogen concentration, a single-neuron adaptive PID controller is used for tracking control. The dissolved oxygen concentration and nitrate nitrogen concentration are respectively controlled by the change ΔK of the oxygen transfer coefficient in the 5th zone. L The changes in a5(t) and internal return flow rate ΔQ a(t) is adjusted. For the single-neuron adaptive control method, this invention uses a supervised Delta learning rule to design an incremental controller, which is defined as follows:

[0117]

[0118] K1 is the neuron proportionality coefficient, satisfying K1>0. i The connection weight coefficients are calculated using the following formula (control formula):

[0119]

[0120] Where, γ i >0 represents the learning rate of the neuron, v1(t), v2(t), and v3(t) represent the inputs of the neuron, and Δu(t) = [ΔK] L a5(t),ΔQ a (t)] T ΔK L a5(t) is the change in the oxygen transfer coefficient of the fifth region at time t, ΔQ a e(t) is the change in return flow rate at time t, and e(t) = z * (t)-z(t) represents the control error between the actual output and the optimized setpoint, z * (t)=[S O * (t),S NO * [(t)] represents the optimal setpoints for dissolved oxygen concentration and nitrate nitrogen concentration at time t, z(t) = [S O (t),S NO [(t)] represents the actual output dissolved oxygen concentration and nitrate nitrogen concentration at time t.

[0121] Specifically, the output of the neural network-based dynamic multi-objective optimization control system for wastewater treatment is the actual dissolved oxygen S. O and nitrate nitrogen S NO concentration value, Figure 3 This is a graph showing the setpoint and tracking results of dissolved oxygen concentration. The solid line represents the optimized setpoint for dissolved oxygen concentration, and the dashed line represents the actual dissolved oxygen output concentration. The horizontal axis represents time (days), and the vertical axis represents dissolved oxygen concentration (mg / L). Figure 4 This is a graph showing the error in dissolved oxygen concentration tracking and control. The horizontal axis represents time (days), and the vertical axis represents the error value of dissolved oxygen concentration (mg / L). Figure 5 This is a graph showing the setpoint and tracking results for nitrate nitrogen concentration. The solid line represents the optimized setpoint for nitrate nitrogen concentration, and the dashed line represents the actual output concentration. The horizontal axis represents time (days), and the vertical axis represents nitrate nitrogen concentration (mg / L). Figure 6This is a graph showing the tracking and control error of nitrate nitrogen concentration. The horizontal axis represents time, in days, and the vertical axis represents the error value of nitrate nitrogen concentration, in milligrams per liter.

[0122] The beneficial effects of this invention are as follows:

[0123] (1) Based on the dynamic characteristics and operation data of the sewage treatment process, this invention establishes an energy consumption and effluent water quality model using an echo state network and sets it as the optimization objective function. The optimized set values ​​of dissolved oxygen concentration and nitrate nitrogen concentration are obtained through a dynamic multi-objective optimization algorithm based on two-space prediction. The invention also combines a single neuron adaptive controller to achieve tracking control of dissolved oxygen concentration and nitrate nitrogen concentration, thus solving the problems of difficult effluent water quality and high operating energy consumption.

[0124] (2) The present invention adopts an optimization control method that combines a dynamic multi-objective optimization algorithm based on two-space prediction with a single neuron adaptive controller to achieve optimization and tracking control of the set values ​​of dissolved oxygen concentration and nitrate nitrogen concentration, thereby reducing operating costs while ensuring that the effluent water quality meets the standards.

[0125] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0126] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A dynamic multi-objective optimization control method for wastewater treatment processes based on neural networks, characterized in that, include: The characteristic variables of effluent quality and energy consumption in the wastewater treatment process are obtained; wherein, the characteristic variables include: dissolved oxygen concentration, nitrate nitrogen concentration, suspended solids concentration, ammonia nitrogen concentration and influent flow rate; Based on echo state networks, an effluent quality model and an energy consumption model to be optimized are constructed according to characteristic variables. Based on the dynamic multi-objective optimization algorithm of two-space prediction, the effluent water quality model and the energy consumption model to be optimized are optimized to obtain the target effluent water quality model and the target energy consumption model. Determine the optimal dissolved oxygen concentration and optimal nitrate nitrogen concentration in the target effluent water quality model and the target energy consumption model; Based on the control formula, and according to the optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration, a single neuron adaptive controller is used to dynamically track and control the wastewater treatment process. A dynamic multi-objective optimization algorithm based on two-space prediction is used to optimize the effluent water quality model and the energy consumption model to be optimized, resulting in a target effluent water quality model and a target energy consumption model, including: Set the population size to N∈[50,150] and the maximum number of iterations to G. max ∈[40,50]; Using the energy consumption model and the effluent water quality model to be optimized as optimization objectives, and the dissolved oxygen concentration in the fifth zone and the nitrate nitrogen concentration in the second zone of the BSM1 model as decision variables, the population is initialized based on the dynamic multi-objective optimization algorithm of two-space prediction, the objective function value is calculated, and the target energy consumption model and the target effluent water quality model are obtained. A dynamic multi-objective optimization algorithm based on two-space prediction is used to initialize the population, calculate the objective function value, and obtain the target energy consumption model and the target effluent water quality model, including: Obtain the environment vector; The environment vector is normalized to obtain a normalized vector; If the Euclidean distance between the normalized vectors is greater than the environmental change threshold, then the optimized setpoints for dissolved oxygen and nitrate nitrogen concentrations are calculated using a dynamic multi-objective optimization algorithm based on two-space prediction to obtain the target energy consumption model and the target effluent water quality model. If not, then the initial population at the next moment is predicted based on the initial population at the current moment and the initial population at the previous moment to obtain the initial population groups at each moment. The initial population groups at each moment are then optimized based on the decomposition multi-objective evolutionary algorithm until the multi-objective evolutionary algorithm reaches the maximum number of iterations to obtain the target energy consumption model and the target effluent water quality model. Based on the initial population at the current time and the initial population at the previous time, the initial population at the next time is predicted, resulting in the initial population at each time step, including: Within the decision space, based on the initial population at the current moment and the initial population at the previous moment, half of the initial population group is obtained; In the target space, based on the initial population at the current moment and the initial population at the previous moment, obtain the other half of the initial population. The first half of the initial seed cluster and the second half of the initial seed cluster are merged to obtain the initial seed clusters at each time point; The initial population of the new environment consists of two parts: first, in the decision space, 50% of the initial population is generated by establishing a denoising autoencoder prediction model; second, in the target space, the desired target vector is mapped back to the decision space by establishing an inverse model to generate the remaining 50% of the initial population.

2. The dynamic multi-objective optimization control method for wastewater treatment process based on neural networks according to claim 1, characterized in that, The energy consumption model is as follows: ； The energy consumption model at time t has the following input variables: ,in, Let be the dissolved oxygen concentration at time t. Let be the concentration of nitrate nitrogen at time t. Let be the concentration of solid suspended matter at time t. Let be the ammonia nitrogen concentration at time t. Let be the inflow rate at time t. It is the sigmoid activation function. Input the weight matrix into the energy consumption model. This is the weight matrix within the energy storage pool in the energy consumption model. This is the output weight matrix in the energy consumption model. This is the internal state matrix in the energy consumption model. Let be the echo state matrix at time t in the energy consumption model. This is the target output matrix in the energy consumption model.

3. The dynamic multi-objective optimization control method for wastewater treatment process based on neural networks according to claim 2, characterized in that, The effluent water quality model is as follows: ； in, The effluent water quality model at time t. Input the weight matrix into the effluent water quality model. This is the weight matrix within the storage tank in the effluent water quality model. To output the weight matrix in the effluent water quality model, This is the internal state matrix in the effluent water quality model. This is the echo state matrix at time t in the effluent water quality model. This is the target output matrix in the effluent water quality model.

4. The dynamic multi-objective optimization control method for wastewater treatment process based on neural networks according to claim 3, characterized in that, The determination of the optimal dissolved oxygen concentration and optimal nitrate nitrogen concentration in the target effluent water quality model and the target energy consumption model includes: Obtain the solution sets of dissolved oxygen concentration and optimal nitrate nitrogen concentration from the target effluent water quality model and the target energy consumption model; Based on the inflection point calculation formula, the inflection points in the solution sets of dissolved oxygen concentration and optimal nitrate nitrogen concentration are calculated; The optimal dissolved oxygen concentration and the optimal nitrate nitrogen concentration are determined based on the inflection point.

5. The dynamic multi-objective optimization control method for wastewater treatment process based on neural networks according to claim 4, characterized in that, The formula for calculating the inflection point is: ； The solution with the largest value is the inflection point, where, , , The coefficients are determined by the two boundary points in the non-dominated solution set.

6. The dynamic multi-objective optimization control method for wastewater treatment process based on neural networks according to claim 5, characterized in that, The single-neuron adaptive controller is: ； The neuron proportionality coefficient satisfies , These are the weighting coefficients.

7. The dynamic multi-objective optimization control method for wastewater treatment process based on neural networks according to claim 6, characterized in that, The control formula is: ； in, The learning rate of neurons. For the input of neurons, , It is the change in the oxygen transfer coefficient of the fifth region at time t. It is the change in return flow rate at time t. To account for the control error between the actual output and the optimized setpoint, These are the optimal setpoints for dissolved oxygen concentration and nitrate nitrogen concentration at time t. , These are the actual output dissolved oxygen concentration and nitrate nitrogen concentration at time t.

Images