Sewage plant aeration decision-making method, computer equipment and readable storage medium

Abstract

The invention discloses a sewage plant aeration decision-making method and equipment, and the method comprises the steps: obtaining sewage treatment data, and obtaining an aeration quantity prediction value X (t) at a t moment; a blower frequency combination {Ki (N)} for achieving the aeration rate predicted value X (t) is obtained, Ki (N) is the ith blower frequency combination {N1i (t), N2i (t),..., Nni (t)} meeting # imgabs0 #, i = 1, 2,..., Nji (t) is the jth blower frequency, and n is the number of blowers; and acquiring the blower frequency combination Kmin (N) = {N1min (t), N2min (t),..., Nnmin (t)} with the minimum electric energy loss in the blower frequency combination {Ki (N)}, and controlling each blower to work at the corresponding frequency in the blower frequency combination Kmin (N). Efficient control over the aeration process of the sewage treatment plant is achieved through an intelligent optimization algorithm, and the method has the advantages of being rapid in convergence and high in efficiency.

Description

Technical Field

[0001] The present invention relates to an optimization method for aeration decision-making in the field of sewage treatment, and particularly to a sewage treatment plant aeration decision-making method, a computer device, and a readable storage medium, which use reinforcement learning and genetic algorithms to achieve efficient control of the aeration process in a sewage treatment plant. Background Art

[0002] Sewage treatment plants are important facilities for urban environmental protection. Among them, the aeration process is a key step in sewage treatment and has an important impact on treatment effects and energy consumption. In the aeration process, how to effectively optimize decision-making to achieve the goal of both ensuring sewage treatment effects and reducing energy consumption is an important research topic at present.

[0003] Traditional aeration decision optimization methods are usually based on empirical models or mathematical models. However, due to the complexity of the sewage treatment process, these methods often fail to achieve ideal results. Therefore, researchers have begun to try to introduce advanced optimization algorithms for decision optimization. Among them, genetic algorithms have received extensive attention. For example, "A design of higher-level control based genetic algorithms for wastewater treatment plants" written by Hai Trung Do et al. was published in 《Engineering Science and Technology, an International Journal》(Volume 24, Issue 4, 2021, pp. 872-878). Genetic algorithms are a type of global optimization algorithm. It is a search algorithm that simulates natural selection and genetic mechanisms and can find the global optimal solution in complex optimization problems with large scale, non-linearity, multi-peak, and multi-constraints. By simulating operations such as crossover, mutation, and selection in the natural selection process, genetic algorithms can maintain diversity during the search process and avoid the problem of falling into local optimal solutions. This can be referred to in "Optimal design and operation of activated sludge processes: State-of-the-art" written by Rainier Hreiz et al. published in 《Chemical Engineering Journal》(Volume 218, 2015, pp. 900-920). The application of genetic algorithms in the sewage treatment process mainly focuses on two aspects: one is parameter optimization, by optimizing various parameters in the aeration process, such as aeration time, aeration intensity, etc., to achieve the best treatment effect; the other is structural optimization, by optimizing the structure of the sewage treatment plant, such as the size, shape, location of the ponds, etc., to achieve the best energy consumption effect. In recent years, genetic algorithms have achieved some results in the optimization of the sewage treatment process.By optimizing various parameters in the aeration process through genetic algorithms, the goal of ensuring sewage treatment effect while reducing energy consumption is achieved. Reference can be made to "Aeration optimization of a wastewater treatment plant using genetic algorithm" published in "Optimal control applications and methods" (Volume 28, Issue 3, 2007, pp. 191 - 208) by B. Holenda et al. However, there are still some problems with genetic algorithms in the optimization of the sewage treatment process, such as slow convergence speed and easy to fall into local optima. Therefore, how to improve genetic algorithms to adapt to the optimization of the sewage treatment process is an important research direction at present. Reference can be made to "Optimizing Control of Wastewater Treatment Plant with Reinforcement Learning: Technical Evaluation of Twin - Delayed Deep Deterministic Policy Gradient Agent" published in "Transactions on Industrial Informatics" (2024) by KLAWIKOWSKA, Zuzanna and GROCHOWSKI, Micha. Summary of the Invention

[0004] The object of the present invention is to provide a sewage plant aeration decision - making method, a computer device and a readable storage medium, which can achieve efficient control of the aeration process in a sewage treatment plant through intelligent optimization algorithms, and have the characteristics of fast convergence and high efficiency.

[0005] To achieve the above object, the solution of the present invention is as follows:

[0006] A sewage plant aeration decision - making method includes:

[0007] Obtain sewage treatment data to get the predicted aeration volume value X(t) at time t;

[0008] Obtain the blower frequency combination {K i (N)} for realizing the predicted aeration volume value X(t), where K i (N) is the i - th blower frequency combination that satisfies and the blower frequency combination {N1i (t), N 2i (t), …, N ni (t)}, i = 1, 2, …, N ji (t) is the frequency of the j-th blower, and n is the number of blowers;

[0009] Obtain the blower frequency combination {K i (N)} with the minimum power loss among them, which is the blower frequency combination K min (N) = {N 1min (t), N 2min (t), …, N nmin (t)}, and control each blower to work at the corresponding frequency in the blower frequency combination K min (N).

[0010] Among them, obtaining sewage treatment data to get the predicted aeration volume value X(t) at time t includes,

[0011] Construct a reinforcement learning model including a state model S(t), an action model A(t), and a reward mechanism R(t). Among them, the state model S(t) is the environmental state at time t during the sewage treatment process, the action model A(t) is the adjustment amount of the aeration volume under the state S(t), and the reward mechanism R(t) is the water quality performance at time t;

[0012] Construct an aeration volume prediction model,

[0013]

[0014] Among them, η is the learning rate and γ is the discount factor; is the discount factor under the optimal action;

[0015] Use the reinforcement learning algorithm to optimize and solve the aeration volume prediction model to obtain the optimal aeration volume adjustment amount A(t);

[0016] According to the aeration volume at time t - 1 and the aeration volume adjustment amount A(t), obtain the predicted aeration volume value X(t) at time t,

[0017] X(t) = X(t - 1) + A(t)

[0018] Among them, X(t) represents the predicted aeration volume value at time t.

[0019] Among them, the expression of the state model S(t) in the reinforcement learning model is,

[0020]

[0021] Among them, DO(t) is the dissolved oxygen characteristic at time t during the sewage treatment process, The ammonia nitrogen characteristic at time t during the sewage treatment process, T(t) is the temperature at time t during the sewage treatment process, PH(t) is the pH value at time t during the sewage treatment process, C TP (t) is the total phosphorus characteristic of the effluent at time t during the sewage treatment process, J TP (t) is the total phosphorus characteristic of the influent at time t during the sewage treatment process, C TN (t) is the total nitrogen characteristic of the effluent at time t during the sewage treatment process, J TN (t) is the total nitrogen characteristic of the influent at time t during the sewage treatment process, C SS (t) is the suspended solid characteristic of the effluent at time t during the sewage treatment process, J SS (t) is the suspended solid characteristic of the influent at time t during the sewage treatment process.

[0022] Among them, in the state model S(t), the acquisition method of each characteristic is as follows:

[0023] Obtain real-time data during the sewage treatment process;

[0024] Preprocess the real-time data, including filling in missing values and correcting or deleting outliers;

[0025] Calculate the correlation coefficients between all characteristics and the target variable, and select the characteristics with correlation coefficients higher than the set threshold.

[0026] Among them, to obtain the blower frequency combination {K i (N)} with the minimum power loss, the blower frequency combination K min (N) = {N 1min (t), N 2min (t), …, N nmin (t)}, includes:

[0027] Construct a fitness function,

[0028] minE(t) = N 1 (t) + N 2 (t) + … + N n (t)

[0029] St.X(t) ≤ N 1 (t) + N 2 (t) + … + N n (t)

[0030] Among them, minE(t) represents the minimum power loss;

[0031] Use a genetic algorithm to solve and obtain the blower frequency combination that minimizes the power loss.

[0032] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, the steps of a sewage treatment plant aeration decision-making method as described above are implemented.

[0033] A computer-readable storage medium stores a computer program; when the computer program is executed by a processor, the steps of a sewage treatment plant aeration decision-making method as described above are implemented.

[0034] After adopting the above solution, the beneficial effect of the present invention is that by combining the genetic algorithm with reinforcement learning, the aeration process of the sewage treatment plant can be intelligently controlled, and the aeration decision-making process can be optimized. The genetic algorithm quickly finds the best solution by simulating the mechanism of natural selection, and reinforcement learning enables the system to continuously learn and optimize in actual operation, improving the efficiency of the control strategy. This method can not only converge quickly, reduce the calculation time, but also greatly save energy consumption while ensuring water quality and reduce the operating cost. Description of the Drawings

[0035] Figure 1 is the training process of the aeration decision-making model;

[0036] Figure 2 is the comparison between the predicted curve and the actual curve of the aeration volume and water quality parameters based on reinforcement learning;

[0037] Figure 3 is the comparison between the predicted curve and the actual curve of the aeration volume and the aeration strategy;

[0038] Figure 4 is the comparison between the predicted curve and the actual curve of the power consumption and the aeration strategy;

[0039] Figure 5 is the power consumption curve per ton of water for a month of a sewage treatment plant without aeration decision-making;

[0040] Figure 6 is the power consumption curve per ton of water for a month of a sewage treatment plant with aeration decision-making. Detailed Embodiments

[0041] The present invention provides a sewage treatment plant aeration decision-making method, including,

[0042] Obtain sewage treatment data to obtain the predicted aeration volume value X(t) at time t;

[0043] Obtain the blower frequency combination {K i (N)} for realizing the predicted aeration volume value X(t), where K i (N) is the i-th blower frequency combination that satisfies and N1i (t), N 2i (t), …, N ni (t)}, i = 1, 2, …, N ji (t) is the frequency of the j - th blower, and n is the number of blowers;

[0044] Obtain the blower frequency combination {K i (N)} with the minimum power loss, and the blower frequency combination K min (N) = {N 1min (t), N 2min (t), …, N nmin (t)}, and control each blower to work at the corresponding frequency in the blower frequency combination K min (N).

[0045] Among them, obtaining sewage treatment data to get the predicted value X(t) of the aeration volume at time t includes,

[0046] Construct a reinforcement learning model including a state model S(t), an action model A(t), and a reward mechanism R(t). Among them, the state model S(t) is the environmental state at time t during the sewage treatment process, the action model A(t) is the adjustment amount of the aeration volume under the state S(t), and the reward mechanism R(t) is the water quality performance at time t;

[0047] Construct an aeration volume prediction model,

[0048]

[0049] Among them, η is the learning rate and γ is the discount factor; is the discount factor under the optimal action;

[0050] Use the reinforcement learning algorithm to optimize and solve the aeration volume prediction model to obtain the optimal aeration volume adjustment amount A(t);

[0051] According to the aeration volume at time t - 1 and the aeration volume adjustment amount A(t), obtain the predicted value X(t) of the aeration volume at time t,

[0052] X(t) = X(t - 1) + A(t)

[0053] Among them, X(t) represents the predicted value of the aeration volume at time t.

[0054] Among them, the expression of the state model S(t) in the reinforcement learning model is,

[0055]

[0056] Among them, DO(t) is the dissolved oxygen characteristic at time t during the sewage treatment process, The ammonia nitrogen characteristic at time t during the sewage treatment process, T(t) is the temperature at time t during the sewage treatment process, PH(t) is the pH value at time t during the sewage treatment process, C TP (t) is the total phosphorus characteristic of the effluent at time t during the sewage treatment process, J TP (t) is the total phosphorus characteristic of the influent at time t during the sewage treatment process, C TN (t) is the total nitrogen characteristic of the effluent at time t during the sewage treatment process, J TN (t) is the total nitrogen characteristic of the influent at time t during the sewage treatment process, C SS (t) is the suspended solid characteristic of the effluent at time t during the sewage treatment process, J SS (t) is the suspended solid characteristic of the influent at time t during the sewage treatment process.

[0057] Among them, in the state model S(t), the acquisition method of each characteristic is

[0058] Obtain real-time data during the sewage treatment process;

[0059] Preprocess the real-time data, including filling missing values and correcting or deleting outliers;

[0060] Calculate the correlation coefficients between all characteristics and the target variable, and select the characteristics with correlation coefficients higher than the set threshold.

[0061] Among them, to obtain the blower frequency combination {K i (N)} with the minimum power loss, the blower frequency combination K min (N) = {N 1min (t), N 2min (t), …, N nmin (t)}, including

[0062] Construct a fitness function,

[0063] minE(t) = N 1 (t) + N 2 (t) + … + N n (t)

[0064] St.X(t) ≤ N 1 (t) + N 2 (t) + … + N n (t)

[0065] Among them, minE(t) represents the minimum power loss;

[0066] Use the genetic algorithm to solve and obtain the blower frequency combination that minimizes the power loss.

[0067] The technical solutions and beneficial effects of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0068] The embodiments of the present invention include the following steps:

[0069] 1) Data collection and preprocessing, obtaining multiple relevant real-time data from the sewage treatment process, including but not limited to parameters such as dissolved oxygen (DO), ammonia nitrogen temperature, pH value, etc., as well as various indicators of the effluent and influent, such as total phosphorus in the effluent (C TP ), total phosphorus in the influent (J TP ), total nitrogen in the effluent (C TN ), total nitrogen in the influent (J TN ), suspended solids in the effluent (C SS ), suspended solids in the influent (J SS ), ammonia nitrogen in the aerobic zone of the biochemical tank, nitrate nitrogen in the aerobic zone of the biochemical tank, ORP in the anaerobic zone of the biochemical tank, influent flow rate, COD in the effluent, COD in the influent, and dissolved oxygen, etc. It also includes the sludge return flow rate of the biochemical tank and chemical oxygen demand, etc. These data will be used as the basis for model input and optimization objectives, providing an accurate basis for aeration decision-making.

[0070] 2) Feature engineering, performing feature engineering processing on the collected data, including the following specific processes:

[0071] First, preprocess all the original data, including filling missing values, dealing with outliers, etc., to obtain the preprocessed features; among them, first, it is necessary to handle the missing values in the original data, and the missing places can be filled with the mean, median or previous and subsequent data; then deal with the outliers, identify the values that are significantly inconsistent with the overall data distribution, and correct or delete these values;

[0072] Then calculate the correlation coefficients between all features and the target variables (for example, treatment effect or energy consumption), so as to select the features with higher correlation, further screen out the variables that have a greater impact on the model, and thus improve the accuracy and efficiency of the model.

[0073] Through the above steps, features such as dissolved oxygen (DO), ammonia nitrogen total phosphorus in the effluent (C TP ), total phosphorus in the influent (J TP ), total nitrogen in the effluent (C TN ), total nitrogen in the influent (J TN ), suspended solids in the effluent (C SS ), suspended solids in the influent (J SS ) are selected.

[0074] 3) Use the reinforcement learning model to obtain the optimal aeration volume and obtain the predicted value of the aeration volume; including the following specific processes:

[0075] A mathematical model for aeration control is established, and the model form is as follows:

[0076] X(t) = X(t - 1) + △X(t) (1)

[0077] Among them, X(t) represents the predicted value of the aeration volume at time t, X(t - 1) represents the aeration volume at time t - 1, and △X(t) is the aeration volume adjustment amount, which is obtained by the aeration volume adjustment strategy based on reinforcement learning.

[0078] Next, the reinforcement learning algorithm will be used to obtain the aeration volume adjustment amount.

[0079] First, a reinforcement learning model is established to be responsible for the control of the aeration volume. According to the environmental state and water quality target, it adaptively adjusts the aeration volume X(t). A state model S(t), an action model A(t), and a reward mechanism R(t) are introduced into the model.

[0080] A state mathematical model is established, and the model form is as follows:

[0081]

[0082] S(t) represents the environmental state at time t during the sewage treatment process, including dissolved oxygen (DO), ammonia nitrogen total phosphorus in the effluent (C TP ), etc., that is, the features selected in step 2); PH(t) represents the pH value at time t, and T(t) represents the temperature at time t.

[0083] An action mathematical model is established, and the model form is as follows:

[0084] A(t) = △X(t) (3)

[0085] A(t) represents the aeration volume adjustment strategy adopted by the system under the state S(t), that is, the aeration volume adjustment amount △X(t).

[0086] A reward mechanism mathematical model is established, and the model form is as follows:

[0087] R(t) = α · WaterQuality(t) (4)

[0088] Among them, the reward function R(t) is defined based on the water quality performance of the system. α is the weight coefficient, and WaterQuality(t) is the water quality score, which is calculated based on the effluent index. The mathematical model form of WaterQuality(t) is as follows:

[0089]

[0090] Then, the reinforcement learning learns the optimal aeration strategy through the deep Q-network (DQN).

[0091] The Q-value update formula (gas volume demand prediction model) is as follows:

[0092]

[0093] Among them, η is the learning rate, γ is the discount factor, which is used to balance the weights of the current reward and the future reward, and is a constant with a value in the range of [0, 1]; is the discount factor under the optimal action; the Q-value reflects the long-term cumulative reward of taking action A(t) in state S(t). Through multiple iterations, reinforcement learning will continuously optimize the aeration volume, so that the system can give the optimal aeration volume prediction while meeting the water quality requirements.

[0094] Through the above operations, the predicted value X(t) of the aeration volume at time t is obtained, which is also the aeration volume that needs to be achieved for control.

[0095] 4) Obtain the equipment scheduling plan based on the genetic algorithm;

[0096] 41) Construct an optimal equipment scheduling model based on the genetic algorithm. The genetic algorithm is used to optimize the equipment scheduling strategy under the predicted value X(t) of the aeration volume given by reinforcement learning to ensure the lowest power loss. The objective function of the genetic algorithm is:

[0097] Fitness = minE(t) (7)

[0098] Among them, E(t) is the power loss under the current aeration strategy, and the aeration strategy is the current frequency combination of each blower.

[0099] 42) Construct an energy consumption prediction model to describe the relationship between the electricity consumption and the blower frequency. The model form is

[0100] E(t) = f 1 (N 1 (t) + N 2 (t) + … + N n (t)) (8)

[0101] Among them, E represents the power loss; N 1 (t) to N n (t) represent the frequencies of the 1st, 2nd, …, nth blowers at time t, and f 1 represents the relationship between the power loss and the blower frequency.

[0102] Construct an aeration volume prediction model to describe the relationship between the aeration volume and the aeration strategy. The model form is

[0103] X(t) = f 2 (N 1 (t) + N 2 (t) + … + Nn (t)) (9)

[0104] Among them, f 2 represents the relationship between the aeration volume and the blower frequency.

[0105] 43) Search for the optimal equipment scheduling strategy

[0106] Initial population generation: Based on the given aeration volume X(t), multiple equipment scheduling schemes (N 1 (t) + N 2 (t) + … + N n (t)) are randomly generated to form the initial population.

[0107] Fitness evaluation: Use the fitness function to evaluate the fitness value of each individual.

[0108] In the genetic algorithm, fitness evaluation is the process of determining the quality of each individual (i.e., each candidate solution). Each individual represents a potential equipment scheduling strategy or aeration volume adjustment plan. The fitness value is used to guide the genetic algorithm to select better individuals and use them to generate the next generation.

[0109] In this embodiment, it is set that the equipment scheduling scheme N(t) = N 1 (t) + N 2 (t) + … + N n (t), and its fitness is calculated according to the following steps:

[0110] Step a, calculate the energy consumption E(t):

[0111] In this embodiment, N 1 (t) = 10kW, N 2 (t) = 12kW, N 3 (t) = 15kW, then the total energy consumption:

[0112] E(t) = 10 + 12 + 15 = 37kWh

[0113] Step b, calculate the water quality score WaterQuality(t):

[0114] In this embodiment, the effluent water quality is: C COD = 45, C SS = 40, C TP = 0.4, C TN = 8, which meets the water quality standard. Therefore:

[0115] WaterQuality(t) = 1

[0116] Step c, calculate the fitness value:

[0117] In this embodiment, a weighted coefficient λ representing energy consumption is set 1 = 0.7, and a weighted coefficient λ representing water quality score 2 = 0.3. Then, the fitness value is:

[0118] Fitness(X(t)) = 0.7 * 37 + 0.3 * 1 = 26.2

[0119] Step d, compare the fitness of different individuals: After calculating the fitness values for all individuals, the genetic algorithm selects individuals with high fitness (i.e., equipment scheduling schemes with low energy consumption and qualified water quality) as the parents of the next generation for crossover and mutation operations.

[0120] Selection: According to the fitness value, select individuals with high fitness as the parents of the next generation.

[0121] Crossover: Perform crossover operations on the selected parents to generate new individuals. The crossover operation formula is:

[0122] child i = parent i λ + child i (1 - λ)(10)

[0123] where, parent i represents the characteristics or parameter values of the i-th parent individual. In this embodiment, it represents the scheduling strategy of the i-th parent individual, usually referring to the energy consumption or running time of each device in equipment scheduling, etc.; child i represents the characteristics or parameter values of the new individual generated by the crossover operation, which is generated through partial information of the parent individuals; λ represents the crossover coefficient (crossover rate), which is used to control the proportion of the two parent individuals in the crossover process. Its value ranges from 0 to 1, and the higher the value, the more the new individual inherits the characteristics of the corresponding parent individual.

[0124] Mutation: Perform mutation operations on the newly generated individuals to increase the diversity of the population. The mutation operation formula is:

[0125] child > ′ = child > + Δ (11)

[0126] where, child > ′ represents the characteristics or parameter values of the mutated individual; Δ is the mutation amount, which represents the amplitude of the change in the characteristics of the individual. Its size determines the amplitude of the mutation. A larger mutation amount will bring greater changes, but it may destroy the good solutions obtained. A smaller mutation amount is helpful for exploring the solution space in a small range.

[0127] Iterative optimization: Repeatedly perform selection, crossover, and mutation operations until a preset termination condition is met. The termination condition can be reaching the preset maximum number of iterations or the fitness value no longer improving significantly.

[0128] 44) Design a fitness function with power loss and effluent water quality as the optimization objectives. The fitness function is defined as:

[0129] minE(t) = N 1 (t) + N 2 (t) + … + N n (t)

[0130] St.X(t - 1) + △X(t) ≤ N 1 (t) + N 2 (t) + … + N n (t) (12)

[0131] where minE(t) represents minimizing power loss, and the N 1 (t) + N 2 (t) + … + N n (t) The aeration volume under the aeration strategy can meet the optimal aeration volume X(t) provided by reinforcement learning.

[0132] 5) Generate the optimal aeration volume and equipment scheduling strategy: Through the collaborative optimization of reinforcement learning and genetic algorithm, generate two key strategies:

[0133] Optimal aeration volume X(t): Provided by reinforcement learning, ensuring compliance with water quality standards and minimizing energy consumption as much as possible.

[0134] Optimal equipment scheduling strategy: Provided by genetic algorithm, ensuring the optimal equipment combination and operation mode under the given aeration volume, and further reducing energy consumption.

[0135] 10) System implementation: Through the established control system, apply the optimization strategy to the actual sewage treatment plant to achieve intelligent and refined aeration control, ensuring compliance with effluent water quality standards and reducing operating costs. The control system includes a data acquisition module, a prediction model module, an optimization algorithm module, and an execution control module. The overall architecture is as follows:

[0136] 1. Data acquisition module: Real-time collect various parameters during the sewage treatment process.

[0137] 2. Prediction model module: Build a prediction model based on the collected data and obtain the required optimal aeration volume by combining reinforcement learning methods.

[0138] 3. Optimization algorithm module: Continuously optimize the aeration decision using genetic algorithm.

[0139] 4. Execution control module: According to the optimization result, adjust the frequency parameter of the blower in real time.

[0140] The method of the present invention optimizes the aeration decision of the sewage treatment plant through the genetic algorithm, effectively improving the intelligent level of the aeration process, significantly reducing the energy consumption, optimizing the operating cost, and providing an innovative solution for energy conservation and emission reduction in the sewage treatment industry.

[0141] Collect the water quality parameter data and aeration volume from January to September 2024. Use the data from January to July as training data to train the model, and the data from July to September as the test. Figure 2 Show the test results. According to the water quality parameters, predict how much aeration volume is required. Compare the predicted aeration volume with the actual aeration volume. Among them, the R2 score is as high as 0.96, and the mean square error is 8.6, within 10.

[0142] Collect the aeration volume and the frequency parameter of the blower from January to September 2024. Use the data from January to July as training data to train the model, and the data from July to September as the test. Predict how much aeration volume can be provided by the current frequency through the blower frequency. Figure 3 Show the test results. Compare the predicted aeration volume with the actual aeration volume. Among them, the R2 score is as high as 0.96, and the mean square error is 7.3, within 10.

[0143] Collect the power loss of the blower and the frequency parameter of the blower from January to September 2024. Use the data from January to July as training data to train the model, and the data from July to September as the test. Predict how much power is consumed by the current frequency through the blower frequency. Figure 4 Show the test results. Take 300 values as a reference. Compare the predicted aeration volume with the actual aeration volume. Among them, the R2 score is as high as 0.99, and the mean square error is 2.1.

[0144] Refer to again Figure 5 and Figure 6 , Figure 5 is the daily power loss curve of not using the present invention in September 2024, and the daily power consumption per ton of water is 0.265; while Figure 6 is the daily power loss curve of using the decision-making suggestion of the present invention as a reference in October 2024. The daily power consumption per ton of water is 0.238, saving 10.2%, and theoretically 20% of the loss can be saved.

[0145] The embodiment of the present invention also provides another computer device, including a processor and a memory configured to store a computer program that can run on the processor; wherein, when the processor is configured to run the computer program, it executes the method steps in the foregoing embodiment.

[0146] In practical applications, the above-mentioned processor includes a Field-Programmable Gate Array (FPGA). The processor can be a Central Processing Unit (CPU) or a Digital Signal Processor (DSP). It can be understood that for different devices, the electronic devices used to implement the functions of the above-mentioned processor can also be others, and the embodiments of the present invention do not make specific limitations.

[0147] The above-mentioned memory can be a volatile memory, such as a Random-Access Memory (RAM); or a non-volatile memory, such as a Read-Only Memory (ROM), a flash memory, a Hard Disk Drive (HDD), or a Solid-State Drive (SSD); or a combination of the above types of memories, and provides instructions and data to the processor.

[0148] In an exemplary embodiment, the embodiments of the present invention also provide a computer-readable storage medium for storing a computer program.

[0149] Optionally, the computer-readable storage medium can be applied to any one of the methods in the embodiments of the present invention, and the computer program causes the computer to execute the corresponding processes implemented by the processor in each of the methods of the embodiments of the present invention. For the sake of brevity, it will not be repeated here.

[0150] In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the couplings, direct couplings, or communication connections between the displayed or discussed components can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be electrical, mechanical, or other forms.

[0151] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code. The solutions in the embodiments of the present invention can be implemented in various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript, etc.

[0152] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0153] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0154] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0155] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they know the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0156] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations.

Claims

1. A sewage treatment plant aeration decision-making method, characterized by: include, Obtain sewage treatment data and obtain the predicted value X(t) of aeration volume at time t; Obtain the blower frequency combination {K i (N)}, where K i (N) is the i-th satisfying The blower frequency combination {N 1i (t),N 2i (t),…,N ni (t)}, i = 1, 2, ..., N ji (t) is the j-th blower frequency, n is the number of blowers; Get the blower frequency combination {K i The blower frequency combination K with the minimum power loss in (N)} min (N) = {N 1min (t),N 2min (t),…,N nmin (t)}, and control each blower to the blower frequency combination K min (N) corresponds to the frequency of operation.

2. A sewage treatment plant aeration decision method as claimed in claim 1, characterized in that: Obtain sewage treatment data and obtain the predicted value of aeration volume X(t) at time t. include, A reinforcement learning model including a state model S(t), an action model A(t), and a reward mechanism R(t) is constructed, wherein the state model S(t) is the environmental state at time t during sewage treatment, the action model A(t) is the adjustment amount of aeration under state S(t), and the reward mechanism R(t) is the water quality performance at time t. Construct aeration prediction model, Among them, η is the learning rate and γ is the discount factor; is the discount factor under the optimal action; The aeration volume prediction model is optimized and solved by using a reinforcement learning algorithm to obtain the optimal aeration volume adjustment amount A(t); According to the aeration volume at time t-1 and the aeration volume adjustment amount A(t), the aeration volume prediction value X(t) at time t is obtained. X(t)=X(t-1)+A(t) Where X(t) represents the predicted value of aeration volume at time t.

3. A sewage treatment plant aeration decision method as claimed in claim 2, characterized in that: The expression of the state model S(t) in the reinforcement learning model is: Among them, DO(t) is the dissolved oxygen characteristic at time t in the sewage treatment process, is the ammonia nitrogen characteristic at time t in the sewage treatment process, T(t) is the temperature at time t in the sewage treatment process, PH(t) is the pH at time t in the sewage treatment process, C TP (t) is the total phosphorus characteristic of the effluent at time t during the sewage treatment process, J TP (t) is the total phosphorus characteristic of the influent at time t during the sewage treatment process, C TN (t) is the total nitrogen characteristic of the effluent at time t during the sewage treatment process, J TN (t) is the total nitrogen characteristic of the influent at time t during the sewage treatment process, C SS (t) is the suspended solids characteristic of the effluent at time t during the sewage treatment process, J SS (t) is the suspended solids characteristic of the influent at time t during the sewage treatment process.

4. A sewage treatment plant aeration decision method as claimed in claim 3, characterized in that: In the state model S(t), the method for obtaining each feature is: Obtain real-time data on the sewage treatment process; Preprocessing the real-time data, including filling missing values ​​and correcting or deleting outliers; Calculate the correlation coefficients between all features and the target variable, and select features with correlation coefficients higher than the set threshold.

5. A sewage treatment plant aeration decision method as claimed in claim 1, characterized in that: Get the blower frequency combination {K i The blower frequency combination K with the minimum power loss in (N)} min (N) = {N 1min (t),N 2min (t),…,N nmin (t)}, including, Construct a fitness function, minE(t)=N1(t)+N2(t)+…+N n (t) St.X(t)≤N1(t)+N2(t)+…+N n (t) Among them, minE(t) represents the minimum value of electric energy loss; Genetic algorithm is used to solve the problem and obtain the blower frequency combination that minimizes power loss.

6. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor; characterized in that: When the processor executes the computer program, the steps of the sewage treatment plant aeration decision method as described in any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium storing a computer program; characterized in that: When the computer program is executed by a processor, the steps of a sewage treatment plant aeration decision method as described in any one of claims 1 to 5 are implemented.

Images