An intelligent control method and system for sewage treatment based on microbial activity

Abstract

The present application belongs to the technical field of intelligent operation of sewage treatment, and particularly relates to a sewage treatment intelligent regulation and control method and system based on microbial activity. The method comprises the following steps: obtaining a value range of input features, wherein the input features comprise microbial activity data; finding an optimal feature combination of the input features in the value range of the input features, with the goal of meeting the sewage treatment result standard and minimizing the cost; determining a regulation and control parameter according to the optimal feature combination and using the regulation and control parameter for a sewage treatment system; wherein the regulation and control parameter comprises air flow and carbon source dosage; and a cost calculation formula is as follows: cost = (air flow-700) / 100+5×(carbon source dosage-3000) / 1000. The present application can reduce air flow and carbon source dosage under the premise of ensuring that the effluent meets the standard, thereby reducing process cost and having a good application prospect.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation technology for wastewater treatment, specifically relating to an intelligent control method and system for wastewater treatment based on microbial activity. Background Technology

[0002] Wastewater treatment is one of the top 10 carbon-emitting industries. Reducing operating costs (such as energy and chemical consumption) and improving pollutant removal performance are perennial themes in wastewater treatment plant regulation. Currently, the management of system operating parameters in wastewater treatment processes is relatively rudimentary, typically based on manual experience. However, biological wastewater treatment is a nonlinear, time-delayed, and strongly coupled process. Manual management often suffers from lag and uncertainty, making it difficult to meet the requirements for stable, economical, and efficient operation of wastewater treatment systems.

[0003] In biological wastewater treatment systems, microorganisms are the main functional agents, and their metabolic activity plays a decisive role in pollutant treatment performance. Specific oxygen respiration rate (SOUR) and adenosine triphosphate (ATP) are rapid and sensitive indicators. Previous studies have shown that SOUR can characterize the activity of denitrifying microorganisms in activated sludge. As the main carrier of usable energy in cellular metabolism, ATP can characterize the quantity and metabolic activity of microorganisms. Therefore, using SOUR and ATP to characterize the state and metabolic activity of microorganisms and correlating them with pollutant degradation is a promising method for optimizing and controlling wastewater treatment systems.

[0004] Furthermore, due to the powerful nonlinear capabilities of artificial intelligence (AI), utilizing AI methods to solve the problems of early warning and precise operation management in wastewater treatment systems is a feasible solution. It is expected to address the lag and uncertainty inherent in manual control of wastewater treatment systems. Chinese invention patent application "CN116798530A: A Method and Device for Intelligent Prediction of Wastewater Treatment Performance" discloses a method based on SOUR and ATP, using an AI model to predict water treatment effects, which can provide key reference parameters for the control of wastewater treatment systems. However, the technical solution in this document can only predict the wastewater treatment effect under specific parameter conditions and cannot directly control process parameters to meet the requirements of effluent compliance and reduced process costs. Therefore, there is still an urgent need in this field to develop a method that can automatically control wastewater treatment process parameters based on specific input features and algorithms. Summary of the Invention

[0005] To address the problems of existing technologies, this invention provides an intelligent control method and system for microbial activity in wastewater treatment.

[0006] A wastewater treatment intelligent control method based on microbial activity, wherein the value range of input features is obtained, the input features including microbial activity data;

[0007] Within the range of values ​​for the input features, the optimal combination of input features is found with the goal of achieving the wastewater treatment target while minimizing the cost.

[0008] Based on the optimal combination of features, control parameters are determined and applied to the wastewater treatment system;

[0009] The control parameters include: airflow rate and carbon source dosage;

[0010] The cost calculation formula is as follows:

[0011] Cost = (Airflow rate - 700) / 100 + 5 × (Carbon source addition - 3000) / 1000.

[0012] Preferably, the method for finding the optimal combination of input features is the particle swarm optimization algorithm.

[0013] Preferably, the steps include:

[0014] Step 1: Traverse at least one input feature to obtain several combinations of input features;

[0015] Step 2: Input the combined features into the machine learning model, and obtain the predicted values ​​of the wastewater treatment result indicators based on the model output;

[0016] Step 3: Determine the corresponding control parameters based on the combination of features;

[0017] Step 4, calculate the cost of each feature combination:

[0018] Step 5: Based on the calculation results of Step 4, select the best feature combination that meets the target index prediction value and has the lowest cost, and apply the control parameters of the best feature combination calculated in Step 3 to the sewage treatment system.

[0019] Preferably, the machine learning model is a random forest model.

[0020] Preferably, in step 3, the method for determining the control parameter is as follows:

[0021] Method 1: The input features also include air flow rate and / or carbon source dosage, and the values ​​of air flow rate and / or carbon source dosage in the feature combination are directly used as the corresponding control parameters;

[0022] Method 2: The input features do not include air flow rate or carbon source dosage. The air flow rate or carbon source dosage is calculated by taking the values ​​of each input feature in the feature combination. The calculation formula is obtained by fitting historical data.

[0023] Preferably, the microbial activity data includes: aerobic tank SOUR.ATP Anoxic pool SOUR ATP_DB and the anoxic pool SOUR MLSS_DB ;

[0024] In the process of finding the optimal combination of input features, the SOUR of the aerobic tank is... ATP and the anoxic pool SOUR MLSS_DB Make changes;

[0025] aerobic tank SOUR ATP The value range is [15, 32], and the step size of the value change for different feature combinations is 1;

[0026] SOUR (Anoxic Pool) MLSS_DB The value range is [0.7, 1.2], and the value change step for different feature combinations is 0.1.

[0027] Preferably, the input features also include airflow rate and / or carbon source dosage;

[0028] The airflow range is [700m]. 3 / h,3000m 3 In the process of finding the optimal feature combination for the input features, the step size for the value changes of different feature combinations is 200m. 3 / h;

[0029] The carbon source dosage ranges from [3000Kg, 7000Kg]. In the process of finding the optimal feature combination for the input features, the step size for the value change of different feature combinations is 200Kg.

[0030] Preferably, the indicator used to evaluate the compliance of wastewater treatment results is total nitrogen in the effluent (total nitrogen in the effluent is calculated from the total nitrogen in the influent and the predicted total nitrogen removal rate).

[0031] This invention also provides an intelligent control system for wastewater treatment with microbial activity, comprising:

[0032] The input module is used to input data;

[0033] The calculation module is used to calculate the control parameters applied to the wastewater treatment system according to the above-mentioned intelligent control method for wastewater treatment;

[0034] The output module is used to output the control parameters to the wastewater treatment system.

[0035] The present invention also provides a computer-readable storage medium having stored thereon a computer program for implementing the above-described intelligent control method for wastewater treatment.

[0036] In this invention, the abbreviations of terms related to water treatment have the following meanings:

[0037]

[0038]

[0039] This invention provides a method that can predict wastewater treatment effectiveness and calculate wastewater treatment costs using microbial activity data, thereby optimizing the control parameters of the wastewater treatment system. This method uses wastewater quality parameters as input data to automatically provide control parameters that balance wastewater treatment effectiveness and cost reduction, avoiding waste caused by manual, experience-based control and ensuring the stable, economical, and efficient operation of the wastewater treatment process. Therefore, this invention has excellent application prospects.

[0040] Obviously, based on the above description of the present invention, and according to common technical knowledge and conventional methods in the field, various other modifications, substitutions or alterations can be made without departing from the basic technical concept of the present invention.

[0041] The following detailed embodiments further illustrate the above-described content of the present invention. However, this should not be construed as limiting the scope of the present invention to the following examples. All technologies implemented based on the above-described content of the present invention fall within the scope of the present invention. Attached Figure Description

[0042] Figure 1 To compare the operating parameters before and after adjustment according to the method in Example 1;

[0043] Figure 2 To compare the operating parameters before and after adjustment according to the method in Example 2;

[0044] Figure 3 To compare the operating parameters before and after adjustment according to the method in Example 3;

[0045] Figure 4 To compare the operating parameters before and after adjustment according to the method of Comparative Example 1;

[0046] Figure 5 This is a heatmap showing the correlation between operating parameters and microbial activity in Experiment Example 2;

[0047] Figure 6 This is a data fitting exercise between the activity of some microorganisms and the operating parameters in Experiment Example 2;

[0048] Figure 7 This is a data fitting of the activity of some microorganisms and operating parameters in Experiment Example 2. Detailed Implementation

[0049] It should be noted that the algorithms for data acquisition, transmission, storage and processing steps not specifically described in the embodiments, as well as the hardware structures and circuit connections not specifically described, can all be implemented using content already disclosed in the prior art.

[0050] Example 1: Intelligent Control Method and System for Wastewater Treatment Based on Microbial Activity

[0051] The system in this embodiment includes:

[0052] The input module is used to input data; the input data includes water quality-related parameters, datasets used for training the model, etc.

[0053] The calculation module is used to calculate the control parameters applied to the wastewater treatment system through intelligent control methods for wastewater treatment.

[0054] The output module is used to output the control parameters to the wastewater treatment system.

[0055] The method for intelligent control of wastewater treatment using the above system includes the following steps:

[0056] Step 1: Traverse each input feature to obtain several sets of input feature combinations; the input features include microbial activity data. In this embodiment, the microbial activity data is selected from the aerobic tank SOUR. ATP (Aerobic tank: total specific oxygen consumption respiration rate with ATP as biomass), Anoxic tank: SOUR ATP_DB (Specific oxygen respiration rate of heterotrophic bacteria in anoxic ponds using ATP as biomass) and SOUR in anoxic ponds MLSS_DB (Specific oxygen consumption respiration rate of heterotrophic bacteria in anoxic ponds, with MLSS as biomass). When traversing microbial activity data, the parameters are set as follows:

[0057] aerobic tank SOUR ATP The value range is [15, 32], and the step size is 1;

[0058] SOUR (Anoxic Pool) MLSS_DB The value range is [0.7, 1.2], with a step size of 0.1;

[0059] In this embodiment, SOUR ATP_DB The value remains unchanged, only the SOUR in the aerobic tank changes. ATP and the anoxic pool SOUR MLSS_DB Because the latter two are more related to the operating parameters of the aerobic tank airflow and carbon source addition.

[0060] Step 2: Input the combined features into the machine learning model, and obtain the predicted value of the wastewater treatment result index based on the model output. In this embodiment, the machine learning model algorithm is the random forest algorithm, and the predicted index is the TN removal rate. Furthermore, using the TN concentration in the influent water quality data and the predicted TN removal rate, the predicted TN concentration of the treated wastewater can be calculated.

[0061] Step 3: Determine the corresponding control parameters based on the feature combinations; the control parameters include airflow rate and carbon source dosage. In this embodiment, airflow rate and carbon source dosage are not used as input features into the machine learning model. Therefore, it is necessary to calculate the airflow rate and carbon source dosage corresponding to each feature combination based on the microbial activity data. The calculation formula is obtained by fitting historical data. This embodiment provides a set of specific calculation formulas:

[0062] Airflow (unit: m) 3 / h)=9.05+0.00774×aerobic tank SOUR ATP (mgO2 / (nmATP·h));

[0063] Carbon source addition (in kg) = 8521.91 - 4519.99 × SOUR of the anoxic tank MLSS_DB (mgO2 / (gMLSS·h)).

[0064] Step 4: Calculate the cost of each feature combination. The cost calculation formula is as follows:

[0065] Cost = (Airflow rate - 700) / 100 + 5 × (Carbon source addition - 3000) / 1000;

[0066] Step 5: Based on the calculation results of Step 4, select the best feature combination that meets the target index prediction value and has the lowest cost, and apply the control parameters of the best feature combination calculated in Step 3 to the sewage treatment system.

[0067] Example 2: Intelligent Control Method and System for Wastewater Treatment Based on Microbial Activity

[0068] The technical solution in this embodiment is basically the same as that in embodiment 1, except that:

[0069] The input features also include: airflow rate and carbon source dosage, and the parameters of the traversal process include:

[0070] The airflow range is [700m]. 3 / h,3000m 3 / h], step size is 200m 3 / h;

[0071] The carbon source dosage ranges from [3000Kg, 7000Kg], with a step size of 200Kg.

[0072] Furthermore, in step 3, there is no need to calculate the air flow rate and carbon source dosage using formulas; the values ​​of the air flow rate and carbon source dosage in the feature combination can be directly used as the corresponding control parameters.

[0073] Example 3: Intelligent Control Method and System for Wastewater Treatment Based on Microbial Activity

[0074] The technical solution in this embodiment is basically the same as that in Embodiment 2, except that: the input features are input into the random forest model, and the PSO optimization algorithm is used to optimize the operating parameters to obtain the optimal airflow and carbon source addition. In the PSO optimization algorithm, the fitness function (i.e., cost) is formulated as follows:

[0075] Cost = (Airflow rate - 700) / 100 + 5 × (Carbon source addition - 3000) / 1000;

[0076] Other parameter settings not specifically described can be performed using the same parameters as in Example 1 or Example 2.

[0077] Comparative Example 1: Intelligent Control Method and System for Wastewater Treatment Based on Operating Parameters

[0078] The technical solution of this comparative example is similar to that of Example 2, except that the input features of the machine learning model do not include microbial activity data (i.e., they do not include the solute OUR of the aerobic tank). ATP Anoxic pool SOUR ATP_DB and the anoxic pool SOUR MLSS_DB This only includes operational parameter data (i.e., only the MLSS of the anoxic pool, air flow rate, and carbon source dosage are used as input features, and only the air flow rate and carbon source dosage are changed during the optimization process, and only these two operational parameter values ​​are optimized).

[0079] The technical solution of the present invention will be further illustrated by the following experiments.

[0080] Example 1: Comparison of the results of different control methods

[0081] I. Experimental Methods

[0082] This experimental example collected operational data from a wastewater treatment plant over a period of time as a dataset, including the following parameters during actual operation: influent water quality (including TN concentration), effluent water quality (including TN concentration), and operational microbial activity data (i.e., SOUR of the aerobic tank). ATP Anoxic pool SOUR ATP_DB and the anoxic pool SOUR MLSS_DB), and the control parameters during operation (i.e., air flow rate, carbon source addition amount, and anoxic pool MLSS).

[0083] Using the cost calculation formula provided in Example 1, the cost in the actual operation of this dataset is calculated and denoted as the cost before adjustment in the following experimental data presentation.

[0084] Then, the optimal control parameters and the corresponding costs during operation were calculated using the methods of Example 1, Example 2, Example 3 and Comparative Example 1, respectively.

[0085] The performance of the random forest models established in the above embodiments and comparative examples is as follows:

[0086]

[0087] The results above demonstrate that the random forest model established in this experimental example has excellent predictive performance.

[0088] II. Experimental Results

[0089] Comparison of the operational data after adjustment according to the method in Example 1 and the original operational data (before adjustment). Figure 1 As shown. A comparison is made between the operating data after adjustment according to the method of Example 2 and the original operating data (before adjustment). Figure 2 As shown. A comparison is made between the operating data after adjustment according to the method of Example 3 and the original operating data (before adjustment). Figure 3 As shown. The comparison between the operating data after adjustment according to the method in Comparative Example 1 and the original operating data (before adjustment) is shown in the figure. Figure 4 As shown. The data for comparison include: air flow rate, carbon source dosage, effluent TN content, and cost.

[0090] As can be seen, after adjustment according to the methods of Examples 1-3, while ensuring that the TN content of the effluent meets the standard, the air flow rate, carbon source dosage and operating costs (costs can reflect the total operating cost) can be effectively reduced, thereby effectively achieving the goal of reducing the operating cost of the sewage treatment process.

[0091] The total airflow rate, total carbon source dosage used in Examples 1-3 and Comparative Example 1, and the costs calculated using the total airflow rate and total carbon source dosage were further compared. The results are shown in the table below:

[0092]

[0093]

[0094] As can be seen, using both microbial activity and operational parameters as input data results in better reductions in airflow, carbon source dosage, and cost, leading to greater cost reduction. Furthermore, the PSO optimization algorithm demonstrates stronger optimization capabilities than traversal optimization algorithms, achieving even lower costs.

[0095] Analysis of the Relationship between Microbial Activity and Wastewater Treatment Operating Parameters in Experiment Example 2

[0096] Using operational data from a wastewater treatment plant over a period of time as the dataset, the correlation between operating parameters and microbial activity in the wastewater treatment system is as follows: Figure 5 As shown. Among them, ATP and SOUR MLSS It is negatively correlated with operating parameters; SOUR ATP It is positively correlated with operating parameters, and most of the correlations are highly significant.

[0097] To further analyze the intrinsic relationship between microbial activity and operating parameters, highly significant correlations were visualized separately and data fitting was performed to clarify the underlying mechanism between microbial activity and operating parameters.

[0098] aerobic tank SOUR ATP Compared with MLSS in aerobic tank and SOUR in aerobic tank ATP Airflow rate in the aerobic tank, SOUR in the anoxic tank ATP_DB The MLSS, ATP, and DO in the anoxic pool all showed highly significant correlations. Therefore, data fitting was performed, and the results are as follows: Figure 6 As shown. The aerobic tank SOUR... ATP The total microbial activity (SOUR) in the aerobic tank showed a linear positive correlation with the MLSS in the aerobic tank, with a correlation coefficient r of 0.74. The fitted linear equation was y = -2.0522 + 0.0023x. As the MLSS in the aerobic tank increased, the total microbial activity (SOUR) in the aerobic tank also increased. ATP The respiratory activity (SOUR) of denitrifying bacteria in the anoxic pond also increases accordingly. ATP_DB The airflow rate in the anoxic tank also showed a linear positive correlation with the MLSS in the anoxic tank, with a Pearson correlation coefficient r of 0.86, and the fitted linear equation was y = 1.65 + 0.0019x. The airflow rate in the aerobic tank was also positively correlated with the SOUR in the aerobic tank. ATPIt also shows a positive linear correlation with a correlation coefficient of 0.63, and the linear equation is y = 9.05 + 0.00774x. The DO and ATP in the anoxic pool show a non-linear negative correlation. When the DO in the anoxic pool increases from 3.7 to 0.43, the ATP in the anoxic pool remains basically unchanged. In order to save energy and maintain high ATP, it is recommended to keep the DO in the anoxic pool at around 0.3. The fitted r is 0.67, and the fitted non-linear equation is y = 386.12 + 44.51 / (sqrt(2*pi)*0.2x)*exp(-(ln(x / 0.2386))^2 / 0.08).

[0099] In the hypoxic pool, ATP and carbon source dosage showed a linear negative correlation, as shown in the data fitting results. Figure 7 As shown, when the ATP in the anoxic pond is in the range of 350-650 nm, the ATP increases with decreasing carbon source dosage. Therefore, to maintain high ATP levels in the anoxic pond microorganisms, the consumption of chemicals can be appropriately reduced, and the carbon source dosage can be kept as low as possible between 3000-3500 kg / d. The Pearson correlation coefficient r for linear fitting is -0.67, and the fitted linear equation is y = 8452 - 9.18x. The relationship between carbon source dosage and SOUR of denitrifying bacteria in the anoxic pond is also discussed. MLSS_DB There is also a good linear negative correlation between them, with a correlation coefficient r of -0.69, and the fitted linear equation is y = 8521.91 - 4519.99x. The sedimentation ratio SV30 and the aerobic tank SOUR... MLSS It also shows a linear negative correlation with a correlation coefficient r of -0.56, and the fitted equation is y = 1.025 - 0.075x.

[0100] Based on the linear and nonlinear fitting results between microbial activity and operating parameters, it can be found that the SOUR of the aerobic tank... ATP It can effectively characterize the MLSS, air flow rate, and SOUR of the aerobic tank. MLSS It can effectively characterize SV30 and SOUR in anoxic ponds. ATP_DB The MLSS in anoxic ponds can be well characterized, and the ATP in anoxic ponds can be well characterized, as can the DO and carbon source dosage in anoxic ponds, and the SOUR in anoxic ponds. MLSS_DB It can also effectively characterize the amount of carbon source added. Given the intrinsic mechanistic relationship between the aforementioned microbial activity and the operating parameters of the wastewater treatment system, this invention selects the aerobic tank SOUR... ATP Anoxic pool SOUR ATP_DB and the anoxic pool SOUR MLSS_DB Three microbial activity data are used to regulate the wastewater treatment system.

[0101] As can be seen from the above embodiments and experimental examples, the present invention provides a method and system for automatically controlling wastewater treatment process parameters, which can reduce airflow and carbon source dosage while ensuring that the effluent meets standards, thereby reducing process costs. Therefore, the present invention has good application prospects.

Claims

1. A smart control method for wastewater treatment based on microbial activity, characterized in that: Obtain the value range of the input features, which include microbial activity data; the microbial activity data includes: SOUR of the aerobic tank. ATP Anoxic pool SOUR ATP_DB and the anoxic pool SOUR MLSS_DB ; Within the range of values ​​for the input features, aiming to achieve the desired wastewater treatment results while minimizing costs, the optimal combination of input features is found. During the process of finding the optimal combination of input features, the SOUR of the aerobic tank is adjusted. ATP and the anoxic pool SOUR MLSS_DB Make changes; Based on the optimal combination of features, control parameters are determined and applied to the wastewater treatment system; The control parameters include: airflow rate and carbon source dosage; The cost calculation formula is as follows: Cost = (Airflow rate - 700) / 100 + 5 × (Carbon source addition - 3000) / 1000; aerobic tank SOUR ATP The value range is [15, 32], and the step size for different feature combinations is 1. SOUR (Anoxic Pool) MLSS_DB The value range is [0.7, 1.2], and the value change step for different feature combinations is 0.1; The input features also include airflow rate and / or carbon source dosage; The airflow range is [700 m 3 / h, 3000 m 3 In the process of finding the optimal feature combination for the input features, the step size for the value changes of different feature combinations is 200 m. 3 / h; The carbon source dosage ranges from [3000 Kg, 7000 Kg]. In the process of finding the optimal feature combination for the input features, the step size for the value change of different feature combinations is 200 Kg. The indicator used to evaluate whether wastewater treatment results meet the standards is total nitrogen in the effluent.

2. The intelligent control method for wastewater treatment according to claim 1, characterized in that: The method for finding the optimal combination of input features is the particle swarm optimization algorithm.

3. The intelligent control method for wastewater treatment according to claim 1, characterized in that, Includes the following steps: Step 1: Traverse at least one input feature to obtain several combinations of input features; Step 2: Input the combined features into the machine learning model, and obtain the predicted values ​​of the wastewater treatment result indicators based on the model output; Step 3: Determine the corresponding control parameters based on the combination of features; Step 4, calculate the cost of each feature combination: Step 5: Based on the calculation results of Step 4, select the best feature combination that meets the target index prediction value and has the lowest cost, and apply the control parameters of the best feature combination calculated in Step 3 to the sewage treatment system.

4. The intelligent control method for wastewater treatment according to claim 3, characterized in that: The machine learning model is a random forest model.

5. The intelligent control method for wastewater treatment according to claim 3, characterized in that: In step 3, the method for determining the control parameter is as follows: Method 1: The input features also include air flow rate and / or carbon source dosage, and the values ​​of air flow rate and / or carbon source dosage in the feature combination are directly used as the corresponding control parameters; Method 2: The input features do not include air flow rate or carbon source dosage. The air flow rate or carbon source dosage is calculated by taking the values ​​of each input feature in the feature combination. The calculation formula is obtained by fitting historical data.

6. A smart control system for wastewater treatment with microbial activity, characterized in that, include: The input module is used to input data; The calculation module is used to calculate the control parameters applied to the wastewater treatment system according to the intelligent control method for wastewater treatment as described in any one of claims 1-5; The output module is used to output the control parameters to the wastewater treatment system.

7. A computer-readable storage medium, characterized in that, It stores a computer program for implementing the intelligent control method for wastewater treatment as described in any one of claims 1-5.

Images